YOLO_V8推理和模型格式转换

1、推理

设置好task、mode、model和测试图片路径source即可。

task: "detect" # choices=['detect', 'segment', 'classify', 'init'] # init is a special case. Specify task to run.
mode: "predict" # choices=['train', 'val', 'predict', 'export'] # mode to run task in.
model: E:\\DLTest\\YOLOv8\\runs\\detect\\best.pt 
source: MaskDataSet/test/images/ # source directory for images or videos

然后执行:\YOLOv8\ultralytics\yolo\v8\detect\predict.py即可输出模型推理结果。

2、模型格式转换

此处以导出ONNX格式为例,主要的几个参数为:

task: "detect" # choices=['detect', 'segment', 'classify', 'init'] # init is a special case. Specify task to run.
mode: "export" # choices=['train', 'val', 'predict', 'export'] # mode to run task in.
model: E:\\DLTest\\YOLOv8\\runs\\detect\\best.pt
format: onnx #torchscript # format to export to
opset: 12  # ONNX: opset version

然后执行:\YOLOv8\ultralytics\yolo\engine\exporter.py 即可输出格式转换之后的模型。

导出的onnx模型和.pt格式的模型在同一路径下。

可能出现的错误:

错误:ONNX: export failure 0.1s: Unsupported ONNX opset version: 17

原因:onnx opset version 设置版本不对,修改版本即可:opset: 12 # ONNX: opset version

猜你喜欢

转载自blog.csdn.net/duan19920101/article/details/128739800