https://github.com/PaddlePaddle/PaddleClas/blob/release/2.6/docs/zh_CN/models/PULC/PULC_safety_helmet.md
python3 -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple pip3 install paddleclas
git clone https://github.com/paddlepaddle/paddleclas -b develop cd paddleclas pip install .
https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.8.1/docs/paddlex/quick_start.md
安装
- 安装PaddlePaddle
安装PaddleX前请先确保您有基础的Python运行环境。（注：当前支持Python 3.8 ～ Python 3.10下运行，更多Python版本适配中）。
# cpu python -m pip install paddlepaddle==3.0.0b1 -i https://www.paddlepaddle.org.cn/packages/stable/cpu/ # gpu，该命令仅适用于 CUDA 版本为 11.8 的机器环境 python -m pip install paddlepaddle-gpu==3.0.0b1 -i https://www.paddlepaddle.org.cn/packages/stable/cu118/ # gpu，该命令仅适用于 CUDA 版本为 12.3 的机器环境 python -m pip install paddlepaddle-gpu==3.0.0b1 -i https://www.paddlepaddle.org.cn/packages/stable/cu123/
- 安装PaddleX
pip install https://paddle-model-ecology.bj.bcebos.com/paddlex/whl/paddlex-3.0.0b1-py3-none-any.whl
一行命令即可快速体验产线效果，
一行命令即可快速体验产线效果，统一的命令行格式为：
paddlex --pipeline [产线名称] --input [输入图片] --device [运行设备]
paddlex --pipeline object_detection --input https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png --device gpu:0 paddlex --pipeline object_detection --input https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png --device cpu
几行代码即可完成产线的快速推理，以小目标检测产线为例：
# https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.8.1/docs/paddlex/quick_start.md
from paddlex import create_pipeline
'''
object_detection 通用目标检测
instance_segmentation 通用实例分割
small_object_detection 小目标检测
'''
pipeline = create_pipeline(pipeline="object_detection")
output = pipeline.predict("mao.png")
for res in output:
res.print() ## 打印预测的结构化输出
res.save_to_img("") ## 保存结果可视化图像
res.save_to_json("") ## 保存预测的结构化输出
{
"input_path": "mao.png",
"boxes": [
{
"cls_id": 0,
"label": "person",
"score": 0.9462200403213501,
"coordinate": [
133.7446746826172,
467.9471740722656,
1767.749267578125,
2047.337646484375
]
},
{
"cls_id": 0,
"label": "person",
"score": 0.8033037781715393,
"coordinate": [
85.89863586425781,
999.6807861328125,
291.049560546875,
1494.3759765625
]
},
{
"cls_id": 0,
"label": "person",
"score": 0.7918640375137329,
"coordinate": [
397.83929443359375,
877.9046630859375,
794.94775390625,
1552.7613525390625
]
},
{
"cls_id": 0,
"label": "person",
"score": 0.699224054813385,
"coordinate": [
1724.633544921875,
940.7135620117188,
1930.8173828125,
1609.0716552734375
]
},
{
"cls_id": 0,
"label": "person",
"score": 0.6661423444747925,
"coordinate": [
1399.666748046875,
1009.642578125,
1623.57763671875,
1440.2664794921875
]
},
{
"cls_id": 0,
"label": "person",
"score": 0.5388690829277039,
"coordinate": [
1501.3985595703125,
967.2034912109375,
1599.070068359375,
1109.2591552734375
]
}
]
}
2 代码测试
安装python
https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.8.1/docs/tutorials/INSTALL.md
# CUDA10.2 python -m pip install paddlepaddle-gpu==2.3.2 -i https://pypi.tuna.tsinghua.edu.cn/simple # CPU python -m pip install paddlepaddle==2.3.2 -i https://pypi.tuna.tsinghua.edu.cn/simple
测试
# check >>> import paddle >>> paddle.utils.run_check() # confirm the paddle's version python -c "import paddle; print(paddle.__version__)"
Install PaddleDetection
# Clone PaddleDetection repository cd <path/to/clone/PaddleDetection> git clone https://github.com/PaddlePaddle/PaddleDetection.git # Install other dependencies cd PaddleDetection pip install -r requirements.txt # Compile and install paddledet python setup.py install
测试
# Predict an image by GPU export CUDA_VISIBLE_DEVICES=0 python tools/infer.py -c configs/ppyolo/ppyolo_r50vd_dcn_1x_coco.yml -o use_gpu=true weights=https://paddledet.bj.bcebos.com/models/ppyolo_r50vd_dcn_1x_coco.pdparams --infer_img=demo/000000014439.jpg
测试代码
import cv2
import paddle
from ppdet.engine import Predictor
from ppdet.config import get_config
import numpy as np
import argparse
'''
python realtime_detect.py --config configs/picodet/application/helmet/picodet_s_416_coco_lcnet.yml --model_dir output_inference
'''
def draw_boxes(img, results, threshold=0.5):
for result in results:
for cls_id, dets in result.items():
for det in dets:
score = det[1]
if score < threshold:
continue
x1, y1, x2, y2 = map(int, det[2:])
label = f"{cls_id}:{score:.2f}"
cv2.rectangle(img, (x1, y1), (x2, y2), (0,255,0), 2)
cv2.putText(img, label, (x1, y1 - 5),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,0), 2)
return img
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--config", type=str, required=True, help="Path to config file.")
parser.add_argument("--model_dir", type=str, required=True, help="Path to inference model.")
args = parser.parse_args()
config="configs/picodet/application/helmet/picodet_s_416_coco_lcnet.yml"
model_dir="output_inference"
# 加载模型配置
cfg = get_config(config, overrides=None)
predictor = Predictor(cfg, model_dir, device='CPU') # 可改为 GPU
cap = cv2.VideoCapture(0) # 打开USB摄像头
while True:
ret, frame = cap.read()
if not ret:
break
results = predictor.predict([frame], visual=False)
frame = draw_boxes(frame, results)
cv2.imshow("Helmet/Intrusion Detection", frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
if __name__ == "__main__":
main()
浙公网安备 33010602011771号