dectron2框架export导出并使用 onnx 记录
python tools/deploy/export_model.py \
--sample-image /Users/gatilin/PycharmProjects/model-graphviz-plot/model_graph/detectron/000000439715.jpg \
--config-file configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml \
--export-method tracing \
--format onnx \
--output ./ \
MODEL.WEIGHTS /Users/gatilin/PycharmProjects/model-graphviz-plot/model_graph/detectron/weights/R-50.pkl \
MODEL.DEVICE cpu
可视化
Talk is cheap. Show me the code