YOLOv5 实例分割 导出ONNX格式

目录

Pre

 1、下载源码，yolov5 v7.0

 2、下载权重

3、配置环境

4、摄像头实例分割预测

5、转ONNX格式

---

Pre

 1、下载源码，yolov5 v7.0

yolov5 只要7.0才有分割，许多人呼吁，作者才把实例分割加入项目中的

 2、下载权重

鼠标左键点击蓝色字就可以，我这里选择YOLOv5n-seg

3、配置环境

pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple

4、摄像头实例分割预测

用 文件进行预测的话，不会弹出窗口显示，只能把视频文件全部预测完成才能查看结果

 python segment/predict.py  --weights yolov5n-seg.pt --sourc 0

CPU推理耗时70ms，与官方60ms接近，数据真实！

别的数据格式也支持推理，比如图片呀

 

5、转ONNX格式

 python export.py --weights yolov5n-seg.pt --include onnx 

输出：

(yolo) jason@honor:~/PycharmProjects/pytorch_learn/yolo/yolov5-7.0$ python export.py --weights yolov5n-seg.pt --include onnx 
export: data=data/coco128.yaml, weights=['yolov5n-seg.pt'], imgsz=[640, 640], batch_size=1, device=cpu, half=False, inplace=False, keras=False, optimize=False, int8=False, dynamic=False, simplify=False, opset=12, verbose=False, workspace=4, nms=False, agnostic_nms=False, topk_per_class=100, topk_all=100, iou_thres=0.45, conf_thres=0.25, include=['onnx']
YOLOv5 🚀 2022-11-22 Python-3.8.13 torch-2.0.0+cu117 CPU

Fusing layers... 
YOLOv5n-seg summary: 224 layers, 1986637 parameters, 0 gradients, 7.1 GFLOPs

PyTorch: starting from yolov5n-seg.pt with output shape (1, 25200, 117) (4.1 MB)

ONNX: starting export with onnx 1.13.1...
============= Diagnostic Run torch.onnx.export version 2.0.0+cu117 =============
verbose: False, log level: Level.ERROR
======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ========================

ONNX: export success ✅ 0.7s, saved as yolov5n-seg.onnx (8.0 MB)

Export complete (1.2s)
Results saved to /home/jason/PycharmProjects/pytorch_learn/yolo/yolov5-7.0
Detect:          python segment/detect.py --weights yolov5n-seg.onnx 
Validate:        python segment/val.py --weights yolov5n-seg.onnx 
PyTorch Hub:     model = torch.hub.load('ultralytics/yolov5', 'custom', 'yolov5n-seg.onnx')  # WARNING ⚠️ SegmentationModel not yet supported for PyTorch Hub AutoShape inference
Visualize:       https://netron.app

转别的格式：

 用netron查看yolov5n-seg.onnx：

 注意，有两个输出。