使用labelme给矿物岩石数据集贴标签

【此文中后段有尝试利用“岩石矿物VS背景” 的手动标注的数据集标签作为训练数据去跑模型，来预测其他未进行手动标注的数据集，但是效果特别特别差，所以只需要参考此文的labelme手动标注岩石矿物数据集的过程，后面如果有训练成功的模型且预测效果还不错会发表出来】

1.1整理原始矿物岩石数据集

1.打开各文件夹内的岩石矿物数据集

在pycharm:检查所有Images是不是RGB三通道(不能是RGBA通道)

（1）先处理为RGB三通道，不然后续labelme标注时会报错

（2）处理为RGB三通道的代码如下

点击查看代码

from PIL import Image
import os

# **************** 您的文件夹路径 ****************
# 注意：在Python字符串中，反斜杠 \ 是转义字符。
# 最好使用原始字符串 r"..." 或将 \ 替换为 / 或 \\
input_folder = r"F:\Dataset\Labelme_images"
# **********************************************

# 确保文件夹存在
if not os.path.isdir(input_folder):
    print(f"错误：文件夹路径不存在或不是目录：{input_folder}")
else:
    print(f"开始处理文件夹：{input_folder}")

    for filename in os.listdir(input_folder):
        # 仅处理常见图像文件类型
        if filename.lower().endswith(('.png', '.jpg', '.jpeg', '.tiff')):
            input_path = os.path.join(input_folder, filename)

            # 跳过目录
            if os.path.isdir(input_path):
                continue

            try:
                img = Image.open(input_path)

                # 检查图像是否为 RGBA 模式
                if img.mode == 'RGBA':
                    print(f"正在转换 {filename}...")

                    # 1. 创建一个白色背景 (R:255, G:255, B:255) 的 RGB 图像
                    rgb_img = Image.new("RGB", img.size, (255, 255, 255))

                    # 2. 将原图粘贴到新的 RGB 图像上，使用 Alpha 通道作为蒙版
                    #    透明部分会与白色背景混合
                    #    img.split()[3] 获取 Alpha (A) 通道
                    rgb_img.paste(img, mask=img.split()[3])

                    # 3. 覆盖原始文件
                    rgb_img.save(input_path)
                    print(f" -> 转换完成，已保存为 RGB 模式。")

                # 检查是否是其他非 RGB/非 RGBA 模式 (如 L=灰度, P=调色板等)
                elif img.mode != 'RGB':
                    print(f"正在转换 {filename} (当前模式: {img.mode})...")
                    # 转换为 RGB 模式
                    rgb_img = img.convert('RGB')
                    rgb_img.save(input_path)
                    print(f" -> 转换完成，已保存为 RGB 模式。")
                else:
                    # 已经是 RGB 模式，跳过
                    print(f" {filename} 已是 RGB 模式，跳过。")


            except Exception as e:
                print(f" -> 处理 {filename} 时发生错误: {e}")

    print("--- 所有文件处理完毕 ---")

（3）输出结果如下：

点击查看代码

D:\anaconda3\envs\labelme\python.exe D:/DeepLearning/Procee_Dataset/ImageType_Convert.py
开始处理文件夹：F:\Label_Dataset\7d489_main\limestone\Images
 000001.jpg 已是 RGB 模式，跳过。
 000002.jpg 已是 RGB 模式，跳过。
......
正在转换 000273.jpg...
 -> 转换完成，已保存为 RGB 模式。
 000274.jpg 已是 RGB 模式，跳过。
......
 000282.jpg 已是 RGB 模式，跳过。
正在转换 000283.jpg...
 -> 转换完成，已保存为 RGB 模式。
 000284.jpg 已是 RGB 模式，跳过。
 000285.jpg 已是 RGB 模式，跳过。
正在转换 000286.jpg...
 -> 转换完成，已保存为 RGB 模式。
 000287.jpg 已是 RGB 模式，跳过。
 ......
 000898.jpg 已是 RGB 模式，跳过。
--- 所有文件处理完毕 ---

进程已结束,退出代码0

（4）处理重复照片

计算图像的感知哈希（phash），对光照/背景变化鲁棒

①手动删除

点击查看代码

import os
import sys
from pathlib import Path
from PIL import Image
import imagehash
import argparse


def get_perceptual_hash(image_path, hash_size=8):
    try:
        with Image.open(image_path) as img:
            if img.mode != 'L':
                img = img.convert('L')
            h = imagehash.phash(img, hash_size=hash_size)
            return str(h)
    except Exception as e:
        print(f"⚠️ 无法处理 {image_path}: {e}")
        return None


def find_and_delete_similar_images(folder_path, hash_size=8, max_distance=5, delete=False):
    image_exts = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp', '.tiff', '.tif', '.jfif'}

    image_files = [
        os.path.join(folder_path, f)
        for f in os.listdir(folder_path)
        if os.path.isfile(os.path.join(folder_path, f)) and Path(f).suffix.lower() in image_exts
    ]

    if not image_files:
        print("未找到任何图片文件。")
        return

    print(f"🔍 扫描到 {len(image_files)} 张图片，正在计算感知哈希...")

    hash_list = []
    valid_files = []

    for f in image_files:
        h_str = get_perceptual_hash(f, hash_size=hash_size)
        if h_str:
            hash_list.append(h_str)
            valid_files.append(f)

    if not hash_list:
        print("❌ 无有效图像可处理。")
        return

    print(f"✅ 成功计算 {len(hash_list)} 张图像的感知哈希。")

    used = [False] * len(valid_files)
    files_to_delete = []

    for i in range(len(valid_files)):
        if used[i]:
            continue
        group = [valid_files[i]]
        used[i] = True
        h1 = imagehash.hex_to_hash(hash_list[i])

        for j in range(i + 1, len(valid_files)):
            if used[j]:
                continue
            h2 = imagehash.hex_to_hash(hash_list[j])
            if (h1 - h2) <= max_distance:
                group.append(valid_files[j])
                used[j] = True

        # 保留 group[0]，其余标记为删除
        if len(group) > 1:
            files_to_delete.extend(group[1:])

    if not files_to_delete:
        print("✅ 未发现冗余图像，无需删除。")
        return

    print(f"🧹 发现 {len(files_to_delete)} 张冗余图像，准备删除...")

    if delete:
        for f in files_to_delete:
            try:
                os.remove(f)
                print(f"🗑️ 已删除: {f}")
            except Exception as e:
                print(f"❌ 删除失败 {f}: {e}")
        print(f"\n✅ 自动删除完成！共删除 {len(files_to_delete)} 张图像。")
    else:
        print("（模拟模式）以下是将被删除的文件：")
        for f in files_to_delete:
            print(f"    ❌ {f}")
        print("\n💡 如需真实删除，请添加 `--delete` 参数运行。")


# ===== 主程序 =====
if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="自动检测并删除视觉相似图像（保留每组第一张）")
    parser.add_argument("folder", nargs="?", help="图像文件夹路径")
    parser.add_argument("--delete", action="store_true", help="启用真实删除（否则仅模拟）")
    parser.add_argument("--hash-size", type=int, default=8)
    parser.add_argument("--max-distance", type=int, default=5)

    args = parser.parse_args()

    try:
        import imagehash
    except ImportError:
        print("❌ 请先安装依赖：pip install imagehash")
        sys.exit(1)

    folder = args.folder or input("请输入图片文件夹的绝对路径: ").strip()
    if not os.path.isdir(folder):
        print("❌ 无效的文件夹路径")
        sys.exit(1)

    print(f"📁 处理文件夹: {folder}")
    print("-" * 50)
    find_and_delete_similar_images(
        folder,
        hash_size=args.hash_size,
        max_distance=args.max_distance,
        delete=args.delete
    )

②自动删除

点击查看代码

import os
import sys
from pathlib import Path
from PIL import Image
import imagehash


def get_perceptual_hash(image_path, hash_size=8):
    try:
        with Image.open(image_path) as img:
            if img.mode != 'L':
                img = img.convert('L')
            h = imagehash.phash(img, hash_size=hash_size)
            return str(h)
    except Exception as e:
        return None


def auto_delete_similar_images(folder_path, hash_size=8, max_distance=5):
    # 支持 .jfif（你提到的格式）
    image_exts = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp', '.tiff', '.tif', '.jfif'}

    image_files = [
        os.path.join(folder_path, f)
        for f in os.listdir(folder_path)
        if os.path.isfile(os.path.join(folder_path, f)) and Path(f).suffix.lower() in image_exts
    ]

    if not image_files:
        print("未找到任何支持的图像文件。")
        return

    print(f"🔍 扫描到 {len(image_files)} 张图像，正在计算感知哈希...")

    hash_list = []
    valid_files = []

    for f in image_files:
        h_str = get_perceptual_hash(f, hash_size=hash_size)
        if h_str is not None:
            hash_list.append(h_str)
            valid_files.append(f)

    print(f"✅ 成功处理 {len(valid_files)} 张有效图像。")

    used = [False] * len(valid_files)
    files_to_delete = []

    for i in range(len(valid_files)):
        if used[i]:
            continue
        h1 = imagehash.hex_to_hash(hash_list[i])
        group = [valid_files[i]]
        used[i] = True

        for j in range(i + 1, len(valid_files)):
            if used[j]:
                continue
            h2 = imagehash.hex_to_hash(hash_list[j])
            if (h1 - h2) <= max_distance:
                group.append(valid_files[j])
                used[j] = True

        if len(group) > 1:
            files_to_delete.extend(group[1:])  # 保留 group[0]，删其余

    if not files_to_delete:
        print("✅ 未发现重复图像。")
        return

    print(f"🗑️ 发现 {len(files_to_delete)} 张重复图像，正在删除...")

    for f in files_to_delete:
        try:
            os.remove(f)
        except Exception as e:
            print(f"❌ 删除失败: {f} - {e}")

    print(f"✅ 自动去重完成！共删除 {len(files_to_delete)} 张图像。")


# ===== 主程序 =====
if __name__ == "__main__":
    try:
        import imagehash
    except ImportError:
        print("❌ 请先运行: pip install imagehash")
        sys.exit(1)

    if len(sys.argv) > 1:
        folder = sys.argv[1]
    else:
        folder = input("请输入图像文件夹路径: ").strip()

    if not os.path.isdir(folder):
        print("❌ 路径无效。")
        sys.exit(1)

    auto_delete_similar_images(folder, hash_size=8, max_distance=5)

1.2在labelme里面标注json文件

1.点击“打开目录”，这样就能同时标注一整个文件夹下的矿物岩石图像

应提前把所有图像放到同一个文件夹中，再点击“打开目录”，用labelme打开这个文件夹标注

2.点击'AI多边形'，手动点击一两下矿物所在区域，labelme就能自动识别出矿物

（1）通过点选想要标注的物体，让AI自动标注，双击，取一个“标签”名，然后Ctrl+S，保存标注的label格式的json文件；如果要切换上一张图像，点击键盘的“D”

【注：在标注过程中，不要任意切换文件夹，否则imagepath会出现问题】

（2）遇到不是单独一个矿物岩石，需要多次点击此岩石并命名为同一个标签名

标注完同一种的多个矿物后，再Ctrl+S保存为json文件

3.打开json文件的imagepath

Labelme标注的json文件最后imagePath图像路径，只能是图像名字本身，不能有其它的路径和符号（例如../、\之类的）

】

1.3在pycharm里面批量将json文件转成mask文件(png格式)

1.先自动改名字，再转为mask文件【提前将json和要改名的jpg文件都放在Labelme_json文件夹里面】

（1）先把Labelme_json文件夹里面的jpg文件例如000459.jpg的所有文件自动转换为类似于dolomite_000459.json命令形式的dolomite_000459.jpg

（2）并在Labelme_mask文件夹里面生成类似于dolomite_000459.png的文件

2.代码如下

点击查看代码

import os
import json
import numpy as np
from PIL import Image
from labelme import utils
import glob

# ====== 路径配置 ======
json_dir = r"G:\Label_Dataset\Dataset\Granite\Labelme_jsons"
output_mask_dir = r"G:\Label_Dataset\Dataset\Granite\Labelme_mask"

# 创建掩膜输出目录（如果不存在）
os.makedirs(output_mask_dir, exist_ok=True)

# 遍历所有 JSON 文件
json_files = glob.glob(os.path.join(json_dir, "*.json"))

# ✅ 修正：添加点号，支持 .webp
img_extensions = ['.jpg', '.jpeg', '.png', '.bmp', '.webp','.gif']

for json_path in json_files:
    json_filename = os.path.basename(json_path)
    print(f"Processing: {json_filename}")

    try:
        # 使用 JSON 文件名（不含扩展名）作为输出基础名
        output_base = os.path.splitext(json_filename)[0]

        # 查找对应图像文件
        img_path = None
        for ext in img_extensions:
            temp_path = os.path.join(json_dir, output_base + ext)
            if os.path.exists(temp_path):
                img_path = temp_path
                break

        if not img_path:
            raise FileNotFoundError(f"未找到对应图像文件（支持格式：{img_extensions}）")

        # 读取 JSON
        with open(json_path, 'r', encoding='utf-8') as f:
            data = json.load(f)

        # 打开图像以获取尺寸（不保存）
        with Image.open(img_path) as img:
            img_width, img_height = img.size

        # 构建标签映射
        label_name_to_value = {"__background__": 0}
        for shape in data["shapes"]:
            label = shape["label"]
            if label not in label_name_to_value:
                label_name_to_value[label] = len(label_name_to_value)

        # 生成掩膜
        lbl, _ = utils.shapes_to_label(
            img_shape=(img_height, img_width),
            shapes=data["shapes"],
            label_name_to_value=label_name_to_value,
        )

        # ✅ 仅保存掩膜（PNG）
        mask_output_path = os.path.join(output_mask_dir, output_base + ".png")
        # 转为 0-255（可选，便于可视化；若用于训练，也可保留 0/1）
        lbl = lbl * 255
        Image.fromarray(lbl.astype(np.uint8), mode='L').save(mask_output_path)

        print(f"✅ 生成掩膜: {output_base}.png")

    except Exception as e:
        print(f"❌ 错误: {json_filename} -> {e}")

print("\n🎉 掩膜生成完成！仅输出 PNG 掩膜文件，文件名与 JSON 一致。")

1.4将mask文件作为“掩码”，直接从原始JPG中提取矿物部分，生成新JPG

1.第一次”掩码“结果：图像全白

【原因：我在labelme标注的json文件转换生成的mask是0-1范围，而非0-255】
labelme的utils.shapes_to_label()默认生成0-1范围的mask（背景=0，矿物=1），不是标准的0-255二值图。

2.将mask范围设为0-255

代码如下：

点击查看代码

from PIL import Image
import numpy as np
import os

# 设置目录路径
original_dir = r"G:\Label_Dataset\Dataset\Granite\Labelme_images"
mask_dir = r"G:\Label_Dataset\Dataset\Granite\Labelme_mask"
output_dir = r"G:\Label_Dataset\Dataset\Granite\Images"

os.makedirs(output_dir, exist_ok=True)

# 支持的原始图像格式（包括 webp）
supported_extensions = ('.jpg', '.jpeg', '.png', '.bmp', '.webp','.gif')

for filename in os.listdir(original_dir):
    if filename.lower().endswith(supported_extensions):
        # 1. 获取基础名（不含扩展名），用于匹配 mask
        base_name, ext = os.path.splitext(filename)
        mask_path = os.path.join(mask_dir, base_name + '.png')

        # 2. 检查 mask 是否存在
        if not os.path.exists(mask_path):
            print(f"⚠️ 警告: 找不到对应的 mask 文件: {mask_path}")
            continue

        # 3. 读取原始图像和 mask
        img = Image.open(os.path.join(original_dir, filename)).convert('RGB')
        mask = Image.open(mask_path).convert('L')

        # 4. 生成二值掩码（>128 视为前景/矿物）
        mask_np = np.array(mask)
        mineral_mask = mask_np > 128  # shape: (H, W), bool

        # 5. 创建白底图像，保留矿物区域
        img_np = np.array(img)
        output_img = np.full_like(img_np, 255)  # 全白背景
        output_img[mineral_mask] = img_np[mineral_mask]

        # 6. 输出为 JPG（无论输入是 webp 还是 jpg）
        output_filename = base_name + '.jpg'
        output_path = os.path.join(output_dir, output_filename)
        Image.fromarray(output_img).save(output_path, "JPEG")

        print(f"✅ 已处理: {filename} → {output_filename}")

1.5划分训练集和测试集(20%测试集，80%训练集)

1.重命名文件：

img_dir：是原始图像
ann_dir：是json转换成的mask的png文件

2.删除文件：

原始json文件删掉

3.提前准备：

将Dolomite、Limestone、Shale三个文件夹下的labelme_images统一放到img_dir文件夹里，将labelme_mask的png文件统一放到ann_dir文件夹下，然后在jupyter notebook打开这个目录： "D:\DeepLearning\Learn_SAM\Label2Everything_labelme2mask\Dataset_DolomiteLimestoneShale"

4.划分训练

打开"D:\DeepLearning\Learn_SAM\Label2Everything_labelme2mask\【C】划分训练-测试集.ipynb"

通过在jupyter notebook里运行该代码，实现80%训练集和20%测试集的划分

5.实现效果如下：

完全可以用这184张 dolomite 图像训练的模型，去为其他几千张不同岩性（limestone, mudstone, sandstone, shale 等）图像生成 mask！

1.6训练模型

1.准备数据：

我的实际数据准备如下：
Dataset_DolomiteLimestoneShale/
├── train/
│ ├── img_dir/ # Dolomite、Limestone和Shale训练图像
│ └ann_dir/ # 对应mask
├── val/
│ ├── img_dir/ # Dolomite、Limestone和Shale验证图像
│ └ann_dir/ # 对应mask
└── Images/ # Mudstone的202张待预测图像

2.训练 U-Net 模型（推荐使用 Keras + TensorFlow）

（1）在Google Colab训练模型，
代码如下：

点击查看代码

import os
import numpy as np
import tensorflow as tf
from tensorflow.keras import layers, models
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from PIL import Image

from google.colab import files

# 上传 train/img_dir 和 train/ann_dir 文件夹（压缩为 .zip）
uploaded = files.upload()

# 解压（假设您上传的是 train.zip 和 val.zip）
!unzip -q train.zip -d /content/
!unzip -q val.zip -d /content/
!unzip -q Images.zip -d /content/

# 查看内容
!ls -R /content/

# 修改为Colab路径
train_img_dir = "/content/train/img_dir"
train_mask_dir = "/content/train/ann_dir"
val_img_dir = "/content/val/img_dir"
val_mask_dir = "/content/val/ann_dir"
new_img_dir = "/content/Images"
output_mask_dir = "/content/Labelme_mask"  # 输出目录

os.makedirs(output_mask_dir, exist_ok=True)

def load_image_and_mask(img_path, mask_path, target_size=(256, 256)):
    try:
        # 加载图像
        img = tf.io.read_file(img_path)
        img = tf.image.decode_image(img, channels=3, expand_animations=False)  # 关闭动画支持
        if img.shape[0] == 0 or img.shape[1] == 0:  # 检查是否为空图像
            raise ValueError(f"图像尺寸为0: {img_path}")

        img = tf.image.resize(img, target_size)
        img = tf.cast(img, tf.float32) / 255.0

        # 加载 mask
        mask = tf.io.read_file(mask_path)
        mask = tf.image.decode_png(mask, channels=1)
        if mask.shape[0] == 0 or mask.shape[1] == 0:
            raise ValueError(f"Mask 尺寸为0: {mask_path}")

        mask = tf.image.resize(mask, target_size)
        mask = tf.cast(mask, tf.float32) / 255.0

        return img, mask

    except Exception as e:
        print(f"❌ 加载失败: {img_path} 或 {mask_path}")
        print(f"   错误信息: {str(e)}")
        # 返回一个占位符（避免中断训练），实际中建议跳过或修复该文件
        dummy_img = tf.zeros((target_size[0], target_size[1], 3), dtype=tf.float32)
        dummy_mask = tf.zeros((target_size[0], target_size[1], 1), dtype=tf.float32)
        return dummy_img, dummy_mask

def create_dataset(image_dir, mask_dir, batch_size=8):
    image_files = [f for f in os.listdir(image_dir) if f.lower().endswith(('.jpg', '.jpeg', '.png'))]
    mask_files = [f for f in os.listdir(mask_dir) if f.lower().endswith('.png')]

    # 按文件名排序
    image_files.sort()
    mask_files.sort()

    # 打印前几个文件名，确认是否匹配
    print("📊 图像文件 (前5个):", image_files[:5])
    print("📊 Mask 文件 (前5个):", mask_files[:5])

    # 确保数量一致
    if len(image_files) != len(mask_files):
        raise ValueError(f"图像和 mask 数量不匹配：{len(image_files)} vs {len(mask_files)}")

    # 检查文件名是否能一一对应（忽略后缀）
    image_basenames = [os.path.splitext(f)[0] for f in image_files]
    mask_basenames = [os.path.splitext(f)[0] for f in mask_files]

    # 匹配相同 base name 的图像和 mask
    matched_pairs = []
    for img_name in image_basenames:
        if img_name in mask_basenames:
            img_idx = image_basenames.index(img_name)
            mask_idx = mask_basenames.index(img_name)
            matched_pairs.append((image_files[img_idx], mask_files[mask_idx]))

    if len(matched_pairs) == 0:
        raise ValueError("没有找到任何匹配的图像和 mask 对！请检查文件名是否一致。")

    image_paths = [os.path.join(image_dir, p[0]) for p in matched_pairs]
    mask_paths = [os.path.join(mask_dir, p[1]) for p in matched_pairs]

    dataset = tf.data.Dataset.from_tensor_slices((image_paths, mask_paths))
    dataset = dataset.map(lambda x, y: load_image_and_mask(x, y), num_parallel_calls=tf.data.AUTOTUNE)
    dataset = dataset.batch(batch_size).prefetch(tf.data.AUTOTUNE)
    return dataset

def unet_model(input_size=(256, 256, 3)):
    inputs = layers.Input(input_size)

    # Encoder
    c1 = layers.Conv2D(16, 3, activation='relu', padding='same')(inputs)
    c1 = layers.Conv2D(16, 3, activation='relu', padding='same')(c1)
    p1 = layers.MaxPooling2D()(c1)

    c2 = layers.Conv2D(32, 3, activation='relu', padding='same')(p1)
    c2 = layers.Conv2D(32, 3, activation='relu', padding='same')(c2)
    p2 = layers.MaxPooling2D()(c2)

    c3 = layers.Conv2D(64, 3, activation='relu', padding='same')(p2)
    c3 = layers.Conv2D(64, 3, activation='relu', padding='same')(c3)
    p3 = layers.MaxPooling2D()(c3)

    c4 = layers.Conv2D(128, 3, activation='relu', padding='same')(p3)
    c4 = layers.Conv2D(128, 3, activation='relu', padding='same')(c4)
    p4 = layers.MaxPooling2D()(c4)

    c5 = layers.Conv2D(256, 3, activation='relu', padding='same')(p4)
    c5 = layers.Conv2D(256, 3, activation='relu', padding='same')(c5)

    # Decoder
    u6 = layers.Conv2DTranspose(128, 2, strides=2, padding='same')(c5)
    u6 = layers.concatenate([u6, c4])
    c6 = layers.Conv2D(128, 3, activation='relu', padding='same')(u6)
    c6 = layers.Conv2D(128, 3, activation='relu', padding='same')(c6)

    u7 = layers.Conv2DTranspose(64, 2, strides=2, padding='same')(c6)
    u7 = layers.concatenate([u7, c3])
    c7 = layers.Conv2D(64, 3, activation='relu', padding='same')(u7)
    c7 = layers.Conv2D(64, 3, activation='relu', padding='same')(c7)

    u8 = layers.Conv2DTranspose(32, 2, strides=2, padding='same')(c7)
    u8 = layers.concatenate([u8, c2])
    c8 = layers.Conv2D(32, 3, activation='relu', padding='same')(u8)
    c8 = layers.Conv2D(32, 3, activation='relu', padding='same')(c8)

    u9 = layers.Conv2DTranspose(16, 2, strides=2, padding='same')(c8)
    u9 = layers.concatenate([u9, c1])
    c9 = layers.Conv2D(16, 3, activation='relu', padding='same')(u9)
    c9 = layers.Conv2D(16, 3, activation='relu', padding='same')(c9)

    outputs = layers.Conv2D(1, 1, activation='sigmoid')(c9)

    model = models.Model(inputs=[inputs], outputs=[outputs])
    return model

train_dataset = create_dataset(train_img_dir, train_mask_dir, batch_size=8)
val_dataset = create_dataset(val_img_dir, val_mask_dir, batch_size=8)

model = unet_model()
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

print("🚀 开始训练...")
history = model.fit(
    train_dataset,
    validation_data=val_dataset,
    epochs=50,
    verbose=1
)

print("✅ 训练完成！")

print("🔮 开始预测新图像（使用 mudstone_ 前缀 + 原始编号）...")

# 获取所有新图像路径（支持 jpg/jpeg/png）
new_image_paths = sorted([
    os.path.join(new_img_dir, f) 
    for f in os.listdir(new_img_dir) 
    if f.lower().endswith(('.jpg', '.jpeg', '.png'))
])

print(f"📦 共找到 {len(new_image_paths)} 张图像待预测")

for img_path in new_image_paths:
    try:
        # 加载图像
        if img_path.lower().endswith('.png'):
            img = tf.io.read_file(img_path)
            img = tf.image.decode_png(img, channels=3)
        else:
            img = tf.io.read_file(img_path)
            img = tf.image.decode_jpeg(img, channels=3)

        img = tf.image.resize(img, (256, 256))
        img = tf.expand_dims(img, axis=0)
        img = tf.cast(img, tf.float32) / 255.0

        # 预测
        pred_mask = model.predict(img, verbose=0)[0]
        pred_mask = (pred_mask > 0.5).astype(np.uint8) * 255

        # ✅ 提取原始文件名中的数字部分（如 limestone_000456.jpg → 000456）
        base_name = os.path.basename(img_path)  # 如 "limestone_000456.jpg"
        if '_' in base_name:
            # 按下划线分割，取最后一部分（数字+扩展名）
            number_with_ext = base_name.split('_')[-1]
            # 去掉扩展名
            number_part = os.path.splitext(number_with_ext)[0]
        else:
            # 如果没有下划线，用整个文件名去掉扩展名
            number_part = os.path.splitext(base_name)[0]

        # ✅ 生成新文件名：mudstone_000456.png
        new_mask_name = f"mudstone_{number_part}.png"
        output_path = os.path.join(output_mask_dir, new_mask_name)

        # 保存
        Image.fromarray(pred_mask.squeeze()).save(output_path)

        # 打印进度（可选）
        print(f"✅ 已生成: {new_mask_name}")

    except Exception as e:
        print(f"❌ 预测失败: {img_path} - {e}")

print(f"✅ 所有 mask 已生成，文件名格式：mudstone_XXXXXX.png")

!zip -r mudstone_masks.zip /content/Labelme_mask/
files.download('mudstone_masks.zip')

（2）报错了
报错的根本原因
【我在 Google Colab 上运行，而代码中的 D:\ 是 Windows 本地磁盘路径，在 Colab（Linux 环境）中完全无效。】

（3）在Colab上传我的文件：
即在运行代码之前输入以下代码：

点击查看代码

from google.colab import files

# 上传 train/img_dir 和 train/ann_dir 文件夹（压缩为 .zip）
uploaded = files.upload()

# 解压（假设您上传的是 train.zip 和 val.zip）
!unzip -q train.zip -d /content/
!unzip -q val.zip -d /content/
!unzip -q Images.zip -d /content/

# 查看内容
!ls -R /content/

【提示：请先在本地将 train/, val/, Images/ 分别压缩成 .zip 文件，然后上传。】

（4）Colab会自动弹窗，让我们加载这三个压缩为Zip的文件夹

（5）修改路径为 Colab 路径：

点击查看代码

# 修改为 Colab 路径
train_img_dir = "/content/train/img_dir"
train_mask_dir = "/content/train/ann_dir"
val_img_dir = "/content/val/img_dir"
val_mask_dir = "/content/val/ann_dir"
new_img_dir = "/content/Images"
output_mask_dir = "/content/Labelme_mask"  # 输出目录

os.makedirs(output_mask_dir, exist_ok=True)

（6）它的运行结果如下：
✅ 所有 202 张图像的 mask 已生成，保存在 /content/Labelme_mask

（7）如何将这 202 张 mask 下载到本地电脑？

①我们看到的路径 /content/Labelme_mask 是 Google Colab 的临时存储空间（Linux 系统）。
Colab 无法直接访问我们本地的 D:\ 磁盘。
所有生成的文件都在云端，重启 Colab 后会被清除（除非我们下载或保存到 Google Drive）。

②如何将这 202 张 mask 下载到本地电脑？

打包成 ZIP 并下载：
在 Colab 中运行以下代码，将整个 Labelme_mask 文件夹压缩并下载：

点击查看代码

!zip -r mudstone_masks.zip /content/Labelme_mask/
files.download('mudstone_masks.zip')

（8）修改1.7_2_(1)中的代码：
【上述模型的训练效果非常差，它预测的202张岩石图的mask后的图像和手动标注的岩石图标签几乎每一张都两模两样】
①在Google Colab里面修改运行这段代码，仍然是对每张图像中的岩石主体进行分割，进行岩石 vs 背景（二值分割）；
②我的训练数据包括184张dolomite、339张limestone和109张shale的JPG原始图像和分别在labelme里面手动标注的mask的png图像，现在想预测202张mudstone的岩石图，要求就是对每张图像中的岩石主体进行分割（即“把岩石从背景中抠出来”）；
③时图像可以不resize 到 256x256，保持原始的图像尺寸就行；
④训练模型的图片一共632张，其中80%是训练集，20%是测试集（已经在1.6_4中划分玩训练集和测试集）

3.训练 改进型U-Net 模型

（1）在Colab改用纯 U-Net 模型 + 数据增强 + 原始尺寸支持

点击查看代码

import os
import numpy as np
import tensorflow as tf
from tensorflow.keras import layers, models
from PIL import Image
from google.colab import files

# =============================
# 第1步：上传并解压数据
# =============================
print("📂 请上传 train.zip, val.zip, Images.zip")
uploaded = files.upload()

!unzip -q train.zip -d /content/
!unzip -q val.zip -d /content/
!unzip -q Images.zip -d /content/

# =============================
# 第2步：设置路径（使用您已划分好的数据）
# =============================
train_img_dir = "/content/train/img_dir"
train_mask_dir = "/content/train/ann_dir"
val_img_dir = "/content/val/img_dir"
val_mask_dir = "/content/val/ann_dir"
new_img_dir = "/content/Images"
output_mask_dir = "/content/Labelme_mask"
os.makedirs(output_mask_dir, exist_ok=True)

# =============================
# 第3步：自定义 padding/unpad 函数（保持宽高比）
# =============================
def pad_to_square(image, target_size=512):
    h, w = tf.shape(image)[0], tf.shape(image)[1]
    max_dim = tf.maximum(h, w)
    scale = tf.cast(target_size / max_dim, tf.float32)
    new_h = tf.cast(tf.cast(h, tf.float32) * scale, tf.int32)
    new_w = tf.cast(tf.cast(w, tf.float32) * scale, tf.int32)
    resized = tf.image.resize(image, [new_h, new_w])
    padded = tf.image.resize_with_pad(resized, target_size, target_size, method='nearest')
    return padded, (h, w, new_h, new_w)

def unpad_from_square(padded_mask, orig_h, orig_w, resized_h, resized_w, target_size=512):
    cropped = tf.image.crop_to_bounding_box(
        tf.expand_dims(padded_mask, axis=-1),
        (target_size - resized_h) // 2,
        (target_size - resized_w) // 2,
        resized_h,
        resized_w
    )
    orig_mask = tf.image.resize(cropped, [orig_h, orig_w], method='nearest')
    return orig_mask

# =============================
# 第4步：带数据增强的加载函数
# =============================
def load_image_and_mask(img_path, mask_path, target_size=512, augment=False):
    # 加载图像
    img = tf.io.read_file(img_path)
    img = tf.image.decode_image(img, channels=3, expand_animations=False)
    mask = tf.io.read_file(mask_path)
    mask = tf.image.decode_png(mask, channels=1)

    # 数据增强（仅训练时使用）
    if augment:
        if tf.random.uniform(()) > 0.5:
            img = tf.image.flip_left_right(img)
            mask = tf.image.flip_left_right(mask)
        if tf.random.uniform(()) > 0.5:
            img = tf.image.flip_up_down(img)
            mask = tf.image.flip_up_down(mask)
        if tf.random.uniform(()) > 0.5:
            img = tf.image.random_brightness(img, 0.2)
            img = tf.image.random_contrast(img, 0.8, 1.2)

    # Pad to square
    img_padded, (h, w, new_h, new_w) = pad_to_square(img, target_size)
    mask_padded, _ = pad_to_square(mask, target_size)

    img_padded = tf.cast(img_padded, tf.float32) / 255.0
    mask_padded = tf.cast(mask_padded, tf.float32) / 255.0

    return img_padded, mask_padded

# =============================
# 第5步：创建数据集（训练集带增强，验证集不带）
# =============================
def create_dataset(image_dir, mask_dir, target_size=512, batch_size=4, augment=False):
    image_files = [f for f in os.listdir(image_dir) if f.lower().endswith(('.jpg', '.jpeg', '.png'))]
    mask_files = [f for f in os.listdir(mask_dir) if f.lower().endswith('.png')]
    
    image_files.sort()
    mask_files.sort()

    image_basenames = [os.path.splitext(f)[0] for f in image_files]
    mask_basenames = [os.path.splitext(f)[0] for f in mask_files]
    pairs = []
    for name in image_basenames:
        if name in mask_basenames:
            img_f = image_files[image_basenames.index(name)]
            mask_f = mask_files[mask_basenames.index(name)]
            pairs.append((os.path.join(image_dir, img_f), os.path.join(mask_dir, mask_f)))
    
    def generator():
        for img_path, mask_path in pairs:
            img, mask = load_image_and_mask(img_path, mask_path, target_size, augment)
            yield img, mask

    dataset = tf.data.Dataset.from_generator(
        generator,
        output_signature=(
            tf.TensorSpec(shape=(target_size, target_size, 3), dtype=tf.float32),
            tf.TensorSpec(shape=(target_size, target_size, 1), dtype=tf.float32)
        )
    )
    return dataset.batch(batch_size).prefetch(tf.data.AUTOTUNE)

# =============================
# 第6步：构建更深的 U-Net 模型（5层）
# =============================
def build_unet(input_size=(512, 512, 3)):
    inputs = layers.Input(input_size)
    
    # Encoder
    c1 = layers.Conv2D(64, 3, activation='relu', padding='same')(inputs)
    c1 = layers.Conv2D(64, 3, activation='relu', padding='same')(c1)
    p1 = layers.MaxPooling2D()(c1)
    
    c2 = layers.Conv2D(128, 3, activation='relu', padding='same')(p1)
    c2 = layers.Conv2D(128, 3, activation='relu', padding='same')(c2)
    p2 = layers.MaxPooling2D()(c2)
    
    c3 = layers.Conv2D(256, 3, activation='relu', padding='same')(p2)
    c3 = layers.Conv2D(256, 3, activation='relu', padding='same')(c3)
    p3 = layers.MaxPooling2D()(c3)
    
    c4 = layers.Conv2D(512, 3, activation='relu', padding='same')(p3)
    c4 = layers.Conv2D(512, 3, activation='relu', padding='same')(c4)
    p4 = layers.MaxPooling2D()(c4)
    
    c5 = layers.Conv2D(1024, 3, activation='relu', padding='same')(p4)
    c5 = layers.Conv2D(1024, 3, activation='relu', padding='same')(c5)
    
    # Decoder
    u6 = layers.Conv2DTranspose(512, 2, strides=2, padding='same')(c5)
    u6 = layers.concatenate([u6, c4])
    c6 = layers.Conv2D(512, 3, activation='relu', padding='same')(u6)
    c6 = layers.Conv2D(512, 3, activation='relu', padding='same')(c6)
    
    u7 = layers.Conv2DTranspose(256, 2, strides=2, padding='same')(c6)
    u7 = layers.concatenate([u7, c3])
    c7 = layers.Conv2D(256, 3, activation='relu', padding='same')(u7)
    c7 = layers.Conv2D(256, 3, activation='relu', padding='same')(c7)
    
    u8 = layers.Conv2DTranspose(128, 2, strides=2, padding='same')(c7)
    u8 = layers.concatenate([u8, c2])
    c8 = layers.Conv2D(128, 3, activation='relu', padding='same')(u8)
    c8 = layers.Conv2D(128, 3, activation='relu', padding='same')(c8)
    
    u9 = layers.Conv2DTranspose(64, 2, strides=2, padding='same')(c8)
    u9 = layers.concatenate([u9, c1])
    c9 = layers.Conv2D(64, 3, activation='relu', padding='same')(u9)
    c9 = layers.Conv2D(64, 3, activation='relu', padding='same')(c9)
    
    outputs = layers.Conv2D(1, 1, activation='sigmoid')(c9)
    return models.Model(inputs=[inputs], outputs=[outputs])

# =============================
# 第7步：训练
# =============================
print("🚀 开始训练...")

# 训练集带数据增强，验证集不带
train_dataset = create_dataset(train_img_dir, train_mask_dir, target_size=512, batch_size=4, augment=True)
val_dataset = create_dataset(val_img_dir, val_mask_dir, target_size=512, batch_size=4, augment=False)

model = build_unet(input_size=(512, 512, 3))
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

history = model.fit(train_dataset, validation_data=val_dataset, epochs=100, verbose=1)
print("✅ 训练完成！")

# =============================
# 第8步：预测（保持原始尺寸）
# =============================
new_image_paths = sorted([
    os.path.join(new_img_dir, f) 
    for f in os.listdir(new_img_dir) 
    if f.lower().endswith(('.jpg', '.jpeg', '.png'))
])

for img_path in new_image_paths:
    try:
        # 加载原始图像
        if img_path.lower().endswith('.png'):
            img = tf.io.read_file(img_path)
            img = tf.image.decode_png(img, channels=3)
        else:
            img = tf.io.read_file(img_path)
            img = tf.image.decode_jpeg(img, channels=3)
        orig_h, orig_w = img.shape[0], img.shape[1]

        # Padding 到 512x512
        img_padded, (h, w, new_h, new_w) = pad_to_square(img, 512)
        img_input = tf.cast(img_padded, tf.float32) / 255.0
        img_input = tf.expand_dims(img_input, axis=0)

        # 预测
        pred_mask_padded = model.predict(img_input, verbose=0)[0]
        pred_mask_padded = (pred_mask_padded > 0.5).astype(tf.float32)

        # 恢复到原始尺寸
        pred_mask_orig = unpad_from_square(pred_mask_padded, h, w, new_h, new_w, 512)
        pred_mask_orig = tf.cast(pred_mask_orig * 255, tf.uint8)

        # 保存
        base_name = os.path.basename(img_path)
        number_part = os.path.splitext(base_name)[0]
        out_path = os.path.join(output_mask_dir, f"mudstone_{number_part}.png")
        Image.fromarray(pred_mask_orig.numpy().squeeze()).save(out_path)

    except Exception as e:
        print(f"❌ 预测失败: {img_path} - {e}")

print("✅ 预测完成！")

# 下载结果
!zip -r mudstone_masks.zip /content/Labelme_mask/
files.download('mudstone_masks.zip')

（2）我把以上代码分步在Colab里面使用GPU运行：

①运行到第7步

②第7步的运行过程中的“127”从何而来？

一共有184张dolomite、339张limestone和109张shale的JPG原始图像，

而我的训练集总图像数 ≈ 632 × 80% ≈ 505 张
每个 batch 包含 4 张图像（batch_size=4）
所以总 batch 数 ≈ 505 / 4 ≈ 126.25 → 向上取整为 127
即【127 是训练集的 batch 数量，与预测的 202 张 mudstone 图像无关！】

③运行到Epoch11的时候已经用时23分58秒

④GPU限额，CPU跑不动了

（3）改到pycharm里面运行

①代码

点击查看代码

import os
import numpy as np
import tensorflow as tf
from tensorflow.keras import layers, models
from PIL import Image
import shutil
import zipfile

# =============================
# 第1步：设置本地路径（请根据您的实际路径修改）
# =============================
base_dir = r"D:\DeepLearning\MineralDataset\Dataset_DolomiteLimestoneShale"

train_img_dir = os.path.join(base_dir, "train", "img_dir")
train_mask_dir = os.path.join(base_dir, "train", "ann_dir")
val_img_dir = os.path.join(base_dir, "val", "img_dir")
val_mask_dir = os.path.join(base_dir, "val", "ann_dir")
new_img_dir = os.path.join(base_dir, "Images")
output_mask_dir = os.path.join(base_dir, "Labelme_mask")
os.makedirs(output_mask_dir, exist_ok=True)

print("✅ 路径设置完成")


# =============================
# 第2步：padding/unpad 函数（无需修改）
# =============================
def pad_to_square(image, target_size=512):
    h, w = tf.shape(image)[0], tf.shape(image)[1]
    max_dim = tf.maximum(h, w)
    scale = tf.cast(target_size / max_dim, tf.float32)
    new_h = tf.cast(tf.cast(h, tf.float32) * scale, tf.int32)
    new_w = tf.cast(tf.cast(w, tf.float32) * scale, tf.int32)
    resized = tf.image.resize(image, [new_h, new_w])
    padded = tf.image.resize_with_pad(resized, target_size, target_size, method='nearest')
    return padded, (h, w, new_h, new_w)


def unpad_from_square(padded_mask, orig_h, orig_w, resized_h, resized_w, target_size=512):
    cropped = tf.image.crop_to_bounding_box(
        tf.expand_dims(padded_mask, axis=-1),
        (target_size - resized_h) // 2,
        (target_size - resized_w) // 2,
        resized_h,
        resized_w
    )
    orig_mask = tf.image.resize(cropped, [orig_h, orig_w], method='nearest')
    return orig_mask


# =============================
# 第3步：数据加载函数（无需修改）
# =============================
def load_image_and_mask(img_path, mask_path, target_size=512, augment=False):
    img = tf.io.read_file(img_path)
    img = tf.image.decode_image(img, channels=3, expand_animations=False)
    mask = tf.io.read_file(mask_path)
    mask = tf.image.decode_png(mask, channels=1)

    if augment:
        if tf.random.uniform(()) > 0.5:
            img = tf.image.flip_left_right(img)
            mask = tf.image.flip_left_right(mask)
        if tf.random.uniform(()) > 0.5:
            img = tf.image.flip_up_down(img)
            mask = tf.image.flip_up_down(mask)
        if tf.random.uniform(()) > 0.5:
            img = tf.image.random_brightness(img, 0.2)
            img = tf.image.random_contrast(img, 0.8, 1.2)

    img_padded, (h, w, new_h, new_w) = pad_to_square(img, target_size)
    mask_padded, _ = pad_to_square(mask, target_size)

    img_padded = tf.cast(img_padded, tf.float32) / 255.0
    mask_padded = tf.cast(mask_padded, tf.float32) / 255.0
    return img_padded, mask_padded


# =============================
# 第4步：创建数据集（无需修改）
# =============================
def create_dataset(image_dir, mask_dir, target_size=512, batch_size=4, augment=False):
    image_files = [f for f in os.listdir(image_dir) if f.lower().endswith(('.jpg', '.jpeg', '.png'))]
    mask_files = [f for f in os.listdir(mask_dir) if f.lower().endswith('.png')]

    image_files.sort()
    mask_files.sort()

    image_basenames = [os.path.splitext(f)[0] for f in image_files]
    mask_basenames = [os.path.splitext(f)[0] for f in mask_files]
    pairs = []
    for name in image_basenames:
        if name in mask_basenames:
            img_f = image_files[image_basenames.index(name)]
            mask_f = mask_files[mask_basenames.index(name)]
            pairs.append((os.path.join(image_dir, img_f), os.path.join(mask_dir, mask_f)))

    def generator():
        for img_path, mask_path in pairs:
            img, mask = load_image_and_mask(img_path, mask_path, target_size, augment)
            yield img, mask

    dataset = tf.data.Dataset.from_generator(
        generator,
        output_signature=(
            tf.TensorSpec(shape=(target_size, target_size, 3), dtype=tf.float32),
            tf.TensorSpec(shape=(target_size, target_size, 1), dtype=tf.float32)
        )
    )
    return dataset.batch(batch_size).prefetch(tf.data.AUTOTUNE)


# =============================
# 第5步：U-Net 模型（无需修改）
# =============================
def build_unet(input_size=(512, 512, 3)):
    inputs = layers.Input(input_size)
    c1 = layers.Conv2D(64, 3, activation='relu', padding='same')(inputs)
    c1 = layers.Conv2D(64, 3, activation='relu', padding='same')(c1)
    p1 = layers.MaxPooling2D()(c1)
    c2 = layers.Conv2D(128, 3, activation='relu', padding='same')(p1)
    c2 = layers.Conv2D(128, 3, activation='relu', padding='same')(c2)
    p2 = layers.MaxPooling2D()(c2)
    c3 = layers.Conv2D(256, 3, activation='relu', padding='same')(p2)
    c3 = layers.Conv2D(256, 3, activation='relu', padding='same')(c3)
    p3 = layers.MaxPooling2D()(c3)
    c4 = layers.Conv2D(512, 3, activation='relu', padding='same')(p3)
    c4 = layers.Conv2D(512, 3, activation='relu', padding='same')(c4)
    p4 = layers.MaxPooling2D()(c4)
    c5 = layers.Conv2D(1024, 3, activation='relu', padding='same')(p4)
    c5 = layers.Conv2D(1024, 3, activation='relu', padding='same')(c5)
    u6 = layers.Conv2DTranspose(512, 2, strides=2, padding='same')(c5)
    u6 = layers.concatenate([u6, c4])
    c6 = layers.Conv2D(512, 3, activation='relu', padding='same')(u6)
    c6 = layers.Conv2D(512, 3, activation='relu', padding='same')(c6)
    u7 = layers.Conv2DTranspose(256, 2, strides=2, padding='same')(c6)
    u7 = layers.concatenate([u7, c3])
    c7 = layers.Conv2D(256, 3, activation='relu', padding='same')(u7)
    c7 = layers.Conv2D(256, 3, activation='relu', padding='same')(c7)
    u8 = layers.Conv2DTranspose(128, 2, strides=2, padding='same')(c7)
    u8 = layers.concatenate([u8, c2])
    c8 = layers.Conv2D(128, 3, activation='relu', padding='same')(u8)
    c8 = layers.Conv2D(128, 3, activation='relu', padding='same')(c8)
    u9 = layers.Conv2DTranspose(64, 2, strides=2, padding='same')(c8)
    u9 = layers.concatenate([u9, c1])
    c9 = layers.Conv2D(64, 3, activation='relu', padding='same')(u9)
    c9 = layers.Conv2D(64, 3, activation='relu', padding='same')(c9)
    outputs = layers.Conv2D(1, 1, activation='sigmoid')(c9)
    return models.Model(inputs=[inputs], outputs=[outputs])


# =============================
# 第6步：训练（添加 ModelCheckpoint 避免丢失）
# =============================
print("🚀 开始训练...")

train_dataset = create_dataset(train_img_dir, train_mask_dir, target_size=512, batch_size=4, augment=True)
val_dataset = create_dataset(val_img_dir, val_mask_dir, target_size=512, batch_size=4, augment=False)

model = build_unet(input_size=(512, 512, 3))
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

# ✅ 保存模型（防止训练中断）
checkpoint = tf.keras.callbacks.ModelCheckpoint(
    os.path.join(base_dir, "best_model.h5"),
    save_best_only=True,
    monitor='val_loss'
)

history = model.fit(
    train_dataset,
    validation_data=val_dataset,
    epochs=80,
    verbose=1,
    callbacks=[checkpoint]
)
print("✅ 训练完成！")

# =============================
# 第7步：预测
# =============================
new_image_paths = sorted([
    os.path.join(new_img_dir, f)
    for f in os.listdir(new_img_dir)
    if f.lower().endswith(('.jpg', '.jpeg', '.png'))
])

for img_path in new_image_paths:
    try:
        if img_path.lower().endswith('.png'):
            img = tf.io.read_file(img_path)
            img = tf.image.decode_png(img, channels=3)
        else:
            img = tf.io.read_file(img_path)
            img = tf.image.decode_jpeg(img, channels=3)
        orig_h, orig_w = img.shape[0], img.shape[1]

        img_padded, (h, w, new_h, new_w) = pad_to_square(img, 512)
        img_input = tf.cast(img_padded, tf.float32) / 255.0
        img_input = tf.expand_dims(img_input, axis=0)

        pred_mask_padded = model.predict(img_input, verbose=0)[0]
        pred_mask_padded = (pred_mask_padded > 0.5).astype(tf.float32)

        pred_mask_orig = unpad_from_square(pred_mask_padded, h, w, new_h, new_w, 512)
        pred_mask_orig = tf.cast(pred_mask_orig * 255, tf.uint8)

        base_name = os.path.basename(img_path)
        number_part = os.path.splitext(base_name)[0]
        out_path = os.path.join(output_mask_dir, f"mudstone_{number_part}.png")
        Image.fromarray(pred_mask_orig.numpy().squeeze()).save(out_path)

    except Exception as e:
        print(f"❌ 预测失败: {img_path} - {e}")

print("✅ 预测完成！")

# =============================
# 第8步：打包结果（可选）
# =============================
shutil.make_archive(
    os.path.join(base_dir, "mudstone_masks"),
    'zip',
    output_mask_dir
)
print(f"📦 结果已打包: {os.path.join(base_dir, 'mudstone_masks.zip')}")

②80轮训练效果如下：

③运行结果

点击查看代码

C:\Users\Administrator\.conda\envs\hyspipeline\python.exe H:\MineralDataset\TrainDataset.py 
✅ 路径设置完成
🚀 开始训练...
2025-12-09 16:59:14.448986: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: SSE SSE2 SSE3 SSE4.1 SSE4.2 AVX AVX2 AVX512F AVX512_VNNI AVX512_BF16 AVX_VNNI AMX_TILE AMX_INT8 AMX_BF16 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
Epoch 1/80
    127/Unknown - 1809s 14s/step - loss: 0.5518 - accuracy: 0.6783C:\Users\Administrator\.conda\envs\hyspipeline\lib\site-packages\keras\src\engine\training.py:3000: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.
  saving_api.save_model(
127/127 [==============================] - 1887s 15s/step - loss: 0.5518 - accuracy: 0.6783 - val_loss: 0.5119 - val_accuracy: 0.7723
Epoch 2/80
127/127 [==============================] - 2452s 19s/step - loss: 0.4959 - accuracy: 0.6980 - val_loss: 0.5061 - val_accuracy: 0.6962
Epoch 3/80
127/127 [==============================] - 2500s 20s/step - loss: 0.4781 - accuracy: 0.7317 - val_loss: 0.4683 - val_accuracy: 0.7150
Epoch 4/80
127/127 [==============================] - 2821s 22s/step - loss: 0.4974 - accuracy: 0.7206 - val_loss: 0.4792 - val_accuracy: 0.6990
Epoch 5/80
127/127 [==============================] - 2913s 23s/step - loss: 0.4441 - accuracy: 0.7502 - val_loss: 0.5897 - val_accuracy: 0.6813
Epoch 6/80
127/127 [==============================] - 2686s 21s/step - loss: 0.4022 - accuracy: 0.7981 - val_loss: 0.5683 - val_accuracy: 0.7349
Epoch 7/80
127/127 [==============================] - 1842s 14s/step - loss: 0.3791 - accuracy: 0.8215 - val_loss: 0.5723 - val_accuracy: 0.7339
Epoch 8/80
127/127 [==============================] - 1863s 15s/step - loss: 0.4112 - accuracy: 0.8098 - val_loss: 0.4388 - val_accuracy: 0.7687
Epoch 9/80
127/127 [==============================] - 1826s 14s/step - loss: 0.3623 - accuracy: 0.8326 - val_loss: 0.4023 - val_accuracy: 0.8079
Epoch 10/80
127/127 [==============================] - 1775s 14s/step - loss: 0.3684 - accuracy: 0.8328 - val_loss: 0.4984 - val_accuracy: 0.7599
Epoch 11/80
127/127 [==============================] - 1732s 14s/step - loss: 0.3594 - accuracy: 0.8343 - val_loss: 0.3968 - val_accuracy: 0.8040
Epoch 12/80
127/127 [==============================] - 1885s 15s/step - loss: 0.3575 - accuracy: 0.8353 - val_loss: 0.3823 - val_accuracy: 0.8165
Epoch 13/80
127/127 [==============================] - 1778s 14s/step - loss: 0.3798 - accuracy: 0.8309 - val_loss: 0.4172 - val_accuracy: 0.7865
Epoch 14/80
127/127 [==============================] - 1787s 14s/step - loss: 0.3579 - accuracy: 0.8357 - val_loss: 0.4437 - val_accuracy: 0.7836
Epoch 15/80
127/127 [==============================] - 1783s 14s/step - loss: 0.3593 - accuracy: 0.8349 - val_loss: 0.4318 - val_accuracy: 0.7908
Epoch 16/80
127/127 [==============================] - 1820s 14s/step - loss: 0.3628 - accuracy: 0.8352 - val_loss: 0.4429 - val_accuracy: 0.7932
Epoch 17/80
127/127 [==============================] - 1851s 15s/step - loss: 0.3638 - accuracy: 0.8316 - val_loss: 0.5143 - val_accuracy: 0.7566
Epoch 18/80
127/127 [==============================] - 1739s 14s/step - loss: 0.3588 - accuracy: 0.8387 - val_loss: 0.4449 - val_accuracy: 0.7921
Epoch 19/80
127/127 [==============================] - 1731s 14s/step - loss: 0.3451 - accuracy: 0.8453 - val_loss: 0.3911 - val_accuracy: 0.8140
Epoch 20/80
127/127 [==============================] - 1713s 13s/step - loss: 0.3433 - accuracy: 0.8457 - val_loss: 0.3775 - val_accuracy: 0.8133
Epoch 21/80
127/127 [==============================] - 1711s 13s/step - loss: 0.3511 - accuracy: 0.8379 - val_loss: 0.3933 - val_accuracy: 0.8041
Epoch 22/80
127/127 [==============================] - 1752s 14s/step - loss: 0.3576 - accuracy: 0.8352 - val_loss: 0.4296 - val_accuracy: 0.7856
Epoch 23/80
127/127 [==============================] - 1784s 14s/step - loss: 0.3418 - accuracy: 0.8474 - val_loss: 0.3992 - val_accuracy: 0.8033
Epoch 24/80
127/127 [==============================] - 1785s 14s/step - loss: 0.3529 - accuracy: 0.8387 - val_loss: 0.4965 - val_accuracy: 0.7749
Epoch 25/80
127/127 [==============================] - 1777s 14s/step - loss: 0.3538 - accuracy: 0.8416 - val_loss: 0.3995 - val_accuracy: 0.8099
Epoch 26/80
127/127 [==============================] - 1730s 14s/step - loss: 0.3484 - accuracy: 0.8424 - val_loss: 0.4101 - val_accuracy: 0.7927
Epoch 27/80
127/127 [==============================] - 1717s 14s/step - loss: 0.3558 - accuracy: 0.8314 - val_loss: 0.3730 - val_accuracy: 0.8214
Epoch 28/80
127/127 [==============================] - 1747s 14s/step - loss: 0.3499 - accuracy: 0.8434 - val_loss: 0.3643 - val_accuracy: 0.8161
Epoch 29/80
127/127 [==============================] - 1877s 15s/step - loss: 0.3415 - accuracy: 0.8490 - val_loss: 0.3824 - val_accuracy: 0.8137
Epoch 30/80
127/127 [==============================] - 1821s 14s/step - loss: 0.3400 - accuracy: 0.8457 - val_loss: 0.3709 - val_accuracy: 0.8315
Epoch 31/80
127/127 [==============================] - 1737s 14s/step - loss: 0.3515 - accuracy: 0.8454 - val_loss: 0.4071 - val_accuracy: 0.8137
Epoch 32/80
127/127 [==============================] - 1820s 14s/step - loss: 0.3351 - accuracy: 0.8560 - val_loss: 0.4203 - val_accuracy: 0.7917
Epoch 33/80
127/127 [==============================] - 1858s 15s/step - loss: 0.3325 - accuracy: 0.8499 - val_loss: 0.4276 - val_accuracy: 0.8051
Epoch 34/80
127/127 [==============================] - 1910s 15s/step - loss: 0.3417 - accuracy: 0.8459 - val_loss: 0.3974 - val_accuracy: 0.8107
Epoch 35/80
127/127 [==============================] - 1868s 15s/step - loss: 0.3600 - accuracy: 0.8369 - val_loss: 0.4159 - val_accuracy: 0.7987
Epoch 36/80
127/127 [==============================] - 1776s 14s/step - loss: 0.3383 - accuracy: 0.8521 - val_loss: 0.4221 - val_accuracy: 0.7947
Epoch 37/80
127/127 [==============================] - 1860s 15s/step - loss: 0.3395 - accuracy: 0.8489 - val_loss: 0.4120 - val_accuracy: 0.8166
Epoch 38/80
127/127 [==============================] - 1771s 14s/step - loss: 0.3408 - accuracy: 0.8498 - val_loss: 0.5695 - val_accuracy: 0.7638
Epoch 39/80
127/127 [==============================] - 1949s 15s/step - loss: 0.3400 - accuracy: 0.8495 - val_loss: 0.4302 - val_accuracy: 0.8057
Epoch 40/80
127/127 [==============================] - 1864s 15s/step - loss: 0.3274 - accuracy: 0.8510 - val_loss: 0.3886 - val_accuracy: 0.8206
Epoch 41/80
127/127 [==============================] - 1714s 13s/step - loss: 0.3239 - accuracy: 0.8562 - val_loss: 0.6162 - val_accuracy: 0.7659
Epoch 42/80
127/127 [==============================] - 1725s 14s/step - loss: 0.3435 - accuracy: 0.8519 - val_loss: 0.4203 - val_accuracy: 0.8075
Epoch 43/80
127/127 [==============================] - 1788s 14s/step - loss: 0.3368 - accuracy: 0.8480 - val_loss: 0.5772 - val_accuracy: 0.7765
Epoch 44/80
127/127 [==============================] - 1765s 14s/step - loss: 0.3230 - accuracy: 0.8628 - val_loss: 0.4734 - val_accuracy: 0.7877
Epoch 45/80
127/127 [==============================] - 2543s 20s/step - loss: 0.3224 - accuracy: 0.8593 - val_loss: 0.4336 - val_accuracy: 0.8109
Epoch 46/80
 93/127 [====================>.........] - ETA: 8:09 - loss: 0.3479 - accuracy: 0.8394

127/127 [==============================] - 1852s 15s/step - loss: 0.3389 - accuracy: 0.8464 - val_loss: 0.4326 - val_accuracy: 0.7916
Epoch 47/80
127/127 [==============================] - 1814s 14s/step - loss: 0.3357 - accuracy: 0.8483 - val_loss: 0.4237 - val_accuracy: 0.8116
Epoch 48/80
127/127 [==============================] - 1813s 14s/step - loss: 0.3233 - accuracy: 0.8582 - val_loss: 0.4014 - val_accuracy: 0.8166
Epoch 49/80
127/127 [==============================] - 1931s 15s/step - loss: 0.3161 - accuracy: 0.8610 - val_loss: 0.4627 - val_accuracy: 0.8236
Epoch 50/80
127/127 [==============================] - 1778s 14s/step - loss: 0.3293 - accuracy: 0.8584 - val_loss: 0.4276 - val_accuracy: 0.8009
Epoch 51/80
115/127 [==========================>...] - ETA: 2:41 - loss: 0.3182 - accuracy: 0.8590 
127/127 [==============================] - 1761s 14s/step - loss: 0.3103 - accuracy: 0.8638 - val_loss: 0.4599 - val_accuracy: 0.8145
Epoch 52/80
127/127 [==============================] - 1722s 14s/step - loss: 0.3310 - accuracy: 0.8556 - val_loss: 0.3952 - val_accuracy: 0.8159
Epoch 53/80
127/127 [==============================] - 1742s 14s/step - loss: 0.3282 - accuracy: 0.8522 - val_loss: 0.5324 - val_accuracy: 0.7971
Epoch 54/80
127/127 [==============================] - 1945s 15s/step - loss: 0.3129 - accuracy: 0.8654 - val_loss: 0.4323 - val_accuracy: 0.8180
Epoch 55/80
127/127 [==============================] - 1813s 14s/step - loss: 0.3499 - accuracy: 0.8371 - val_loss: 0.6181 - val_accuracy: 0.8074
Epoch 56/80
127/127 [==============================] - 1802s 14s/step - loss: 0.3443 - accuracy: 0.8533 - val_loss: 0.4440 - val_accuracy: 0.8217
Epoch 57/80
127/127 [==============================] - 1773s 14s/step - loss: 0.3082 - accuracy: 0.8678 - val_loss: 0.7200 - val_accuracy: 0.7910
Epoch 58/80
127/127 [==============================] - 1738s 14s/step - loss: 0.3212 - accuracy: 0.8593 - val_loss: 0.3792 - val_accuracy: 0.8478
Epoch 59/80
127/127 [==============================] - 1834s 14s/step - loss: 0.3004 - accuracy: 0.8714 - val_loss: 0.4644 - val_accuracy: 0.8035
Epoch 60/80
127/127 [==============================] - 1866s 15s/step - loss: 0.3107 - accuracy: 0.8648 - val_loss: 0.4316 - val_accuracy: 0.8223
Epoch 61/80
127/127 [==============================] - 1699s 13s/step - loss: 0.3136 - accuracy: 0.8625 - val_loss: 0.4537 - val_accuracy: 0.8072
Epoch 62/80
127/127 [==============================] - 1687s 13s/step - loss: 0.3157 - accuracy: 0.8589 - val_loss: 0.3525 - val_accuracy: 0.8377
Epoch 63/80
127/127 [==============================] - 1735s 14s/step - loss: 0.3337 - accuracy: 0.8414 - val_loss: 0.4843 - val_accuracy: 0.8193
Epoch 64/80
127/127 [==============================] - 1710s 13s/step - loss: 0.3126 - accuracy: 0.8639 - val_loss: 0.3473 - val_accuracy: 0.8530
Epoch 65/80
127/127 [==============================] - 1762s 14s/step - loss: 0.3295 - accuracy: 0.8276 - val_loss: 0.4351 - val_accuracy: 0.8068
Epoch 66/80
127/127 [==============================] - 1752s 14s/step - loss: 0.2894 - accuracy: 0.8784 - val_loss: 0.3173 - val_accuracy: 0.8755
Epoch 67/80
127/127 [==============================] - 1710s 13s/step - loss: 0.2621 - accuracy: 0.8908 - val_loss: 0.3319 - val_accuracy: 0.8600
Epoch 68/80
127/127 [==============================] - 2041s 16s/step - loss: 0.2551 - accuracy: 0.8946 - val_loss: 0.3067 - val_accuracy: 0.8814
Epoch 69/80
127/127 [==============================] - 1863s 15s/step - loss: 0.2594 - accuracy: 0.8928 - val_loss: 0.4142 - val_accuracy: 0.8292
Epoch 70/80
127/127 [==============================] - 1818s 14s/step - loss: 0.2587 - accuracy: 0.8922 - val_loss: 0.2773 - val_accuracy: 0.8878
Epoch 71/80
127/127 [==============================] - 1911s 15s/step - loss: 0.2538 - accuracy: 0.8965 - val_loss: 0.2710 - val_accuracy: 0.8877
Epoch 72/80
127/127 [==============================] - 1785s 14s/step - loss: 0.2570 - accuracy: 0.8976 - val_loss: 0.2708 - val_accuracy: 0.8724
Epoch 73/80
127/127 [==============================] - 1856s 15s/step - loss: 0.2578 - accuracy: 0.8899 - val_loss: 0.5696 - val_accuracy: 0.8298
Epoch 74/80
127/127 [==============================] - 1810s 14s/step - loss: 0.2714 - accuracy: 0.8891 - val_loss: 0.3212 - val_accuracy: 0.8833
Epoch 75/80
127/127 [==============================] - 1798s 14s/step - loss: 0.2460 - accuracy: 0.8988 - val_loss: 0.2660 - val_accuracy: 0.8921
Epoch 76/80
127/127 [==============================] - 1727s 14s/step - loss: 0.2417 - accuracy: 0.8999 - val_loss: 0.2653 - val_accuracy: 0.8962
Epoch 77/80
127/127 [==============================] - 1826s 14s/step - loss: 0.2385 - accuracy: 0.9030 - val_loss: 0.2551 - val_accuracy: 0.8987
Epoch 78/80
127/127 [==============================] - 1877s 15s/step - loss: 0.2381 - accuracy: 0.9009 - val_loss: 0.2542 - val_accuracy: 0.8940
Epoch 79/80
127/127 [==============================] - 1976s 16s/step - loss: 0.2338 - accuracy: 0.9027 - val_loss: 0.2549 - val_accuracy: 0.8944
Epoch 80/80
127/127 [==============================] - 2040s 16s/step - loss: 0.2386 - accuracy: 0.9011 - val_loss: 0.2715 - val_accuracy: 0.8905
✅ 训练完成！
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_001.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_002.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_003.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_004.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_005.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_006.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_007.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_008.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_009.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_010.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_011.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_012.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_013.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_014.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_015.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_016.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_017.png - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_018.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_019.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_020.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_021.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_022.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_023.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_024.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_025.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_026.png - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_027.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_028.png - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_029.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_030.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_031.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_032.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_033.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_034.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_035.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_036.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_037.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_038.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_039.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_040.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_041.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_042.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_043.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_044.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_045.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_046.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_047.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_048.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_049.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_050.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_051.png - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_052.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_053.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_054.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_055.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_056.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_057.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_058.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_059.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_060.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_061.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_062.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_063.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_064.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_065.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_066.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_067.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_068.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_069.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_070.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_071.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_072.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_073.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_074.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_075.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_076.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_077.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_078.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_079.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_080.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_081.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_082.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_083.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_084.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_085.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_086.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_087.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_088.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_089.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_090.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_091.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_092.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_093.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_094.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_095.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_096.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_097.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_098.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_099.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_100.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_101.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_102.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_103.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_104.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_105.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_106.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_107.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_108.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_109.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_110.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_111.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_112.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_113.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_114.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_115.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_116.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_117.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_118.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_119.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_120.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_121.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_122.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_123.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_124.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_125.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_126.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_127.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_128.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_129.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_130.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_131.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_132.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_133.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_134.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_135.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_136.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_137.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_138.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_139.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_140.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_141.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_142.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_143.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_144.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_145.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_146.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_147.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_148.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_149.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_150.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_151.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_152.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_153.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_154.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_155.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_156.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_157.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_158.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_159.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_160.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_161.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_162.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_163.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_164.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_165.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_166.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_167.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_168.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_169.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_170.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_171.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_172.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_173.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_174.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_175.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_176.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_177.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_178.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_179.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_180.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_181.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_182.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_183.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_184.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_185.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_186.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_187.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_188.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_189.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_190.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_191.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_192.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_193.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_194.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_195.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_196.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_197.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_198.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_199.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_200.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_201.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_202.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_203.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_204.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_205.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_206.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_207.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_208.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_209.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_210.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_211.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_212.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_213.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_214.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_215.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_216.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_217.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_218.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_219.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_220.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_221.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_222.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_223.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_224.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_225.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_226.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_227.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_228.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_229.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_230.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_231.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_232.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_233.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_234.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_235.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_236.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_237.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_238.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_239.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_240.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_241.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_242.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_243.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_244.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_245.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_246.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_247.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_248.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_249.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_250.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_251.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_252.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_253.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_254.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_255.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_256.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_257.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_258.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_259.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_260.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_261.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_262.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_263.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_264.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_265.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_266.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_267.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_268.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_269.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_270.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_271.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_272.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_273.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_274.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_275.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_276.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_277.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_278.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_279.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_280.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_281.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_282.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_283.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_284.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_285.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_286.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_287.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_288.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_289.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_290.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_291.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_292.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_293.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_294.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_295.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_296.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_297.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_298.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_299.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_300.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_301.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_302.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_303.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_304.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_305.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_306.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_307.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_308.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_309.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_310.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_311.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_312.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_313.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_314.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_315.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_316.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_317.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_318.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_319.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_320.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_321.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_322.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_323.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_324.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_325.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_326.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_327.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_328.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_329.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_330.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_331.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_332.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_333.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_334.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_335.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_336.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_337.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_338.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_339.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_340.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_341.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_342.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_343.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_344.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_345.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_346.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_347.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_348.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_349.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_350.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_351.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_352.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_353.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_354.png - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_355.jpg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_356.jpeg - Cannot interpret 'tf.float32' as a data type
❌ 预测失败: H:\MineralDataset\Dataset_DolomiteLimestoneShale\Images\Basalt_357.jpg - Cannot interpret 'tf.float32' as a data type
✅ 预测完成！
📦 结果已打包: H:\MineralDataset\Dataset_DolomiteLimestoneShale\mudstone_masks.zip

进程已结束，退出代码为 0

【预测失败了，唉，我还是labelme手动标注数据集的标签吧】