Closed huihui308 closed 10 months ago
I found there is a error if it did not add '--nms_class_agnostic' in export_onnx.py.
2023-10-13 10:37:51.043 | INFO | __main__:main:76 - args value: Namespace(batch_size=1, conf_thresh=0.1, dynamic=True, exp_file=None, experiment_name=None, fuse_conv_bn=False, img_size=640, input='images', model='rtcdet_p', nms_class_agnostic=False, nms_thresh=0.5, no_onnxsim=False, num_classes=2, opset=11, opts=[], output='output', save_dir='../weights/onnx/', topk=100, weight='./../weights/plate/rtcdet_p/rtcdet_p_bs256_best_2023-09-27_06-09-12.pth') ============================== Model: RTCDET_P ... ============================== Build RTCDET_P ... ============================== Neck: sppf -------------------------------------- Best model infor: Epoch: 299 mAP: 56.5 -------------------------------------- Finished loading model! 2023-10-13 10:37:51.176 | INFO | __main__:main:92 - loading checkpoint done. export_onnx.py:101: FutureWarning: 'torch.onnx._export' is deprecated in version 1.12.0 and will be removed in 2.0. Please use `torch.onnx.export` instead. torch.onnx._export( /home/david/.local/lib/python3.8/site-packages/torch/functional.py:504: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ../aten/src/ATen/native/TensorShape.cpp:3483.) return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined] ============= Diagnostic Run torch.onnx.export version 2.0.1+cu118 ============= verbose: False, log level: Level.ERROR ======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ======================== 2023-10-13 10:37:53.798 | INFO | __main__:main:112 - generated onnx model named ../weights/onnx/11/rtcdet_p.onnx WARNING: The argument `dynamic_input_shape=True` is not needed any more, onnxsim can now support dynamic input shapes natively, please refer to the latest documentation. An error will be raised in the future. WARNING: The argument `input_shapes` is deprecated. Please use `overwrite_input_shapes` and/or `test_input_shapes` instead. An error will be raised in the future. 2023-10-13 10:37:54.552 | INFO | __main__:main:133 - generated simplified onnx model named ../weights/onnx/11/onnxsim/rtcdet_p.onnx (paper) david@david-ubuntu20:tools$ (paper) david@david-ubuntu20:tools$ (paper) david@david-ubuntu20:tools$ (paper) david@david-ubuntu20:tools$ python3 export_onnx.py --model=rtcdet_p --num_classes=2 --dynamic --weight=./../weights/plate/rtcdet_p/rtcdet_p_bs256_best_2023-09-27_06-09-12.pth 2023-10-13 10:38:33.812 | INFO | __main__:main:75 - args value: Namespace(batch_size=1, conf_thresh=0.1, dynamic=True, exp_file=None, experiment_name=None, fuse_conv_bn=False, img_size=640, input='images', model='rtcdet_p', nms_thresh=0.5, no_onnxsim=False, num_classes=2, opset=11, opts=[], output='output', save_dir='../weights/onnx/', topk=100, weight='./../weights/plate/rtcdet_p/rtcdet_p_bs256_best_2023-09-27_06-09-12.pth') ============================== Model: RTCDET_P ... ============================== Build RTCDET_P ... 2023-10-13 10:38:33.812 | ERROR | __main__:<module>:136 - An error has been caught in function '<module>', process 'MainProcess' (158930), thread 'MainThread' (140610050365248): Traceback (most recent call last): > File "export_onnx.py", line 136, in <module> main() └ <function main at 0x7fe17b4addc0> File "export_onnx.py", line 82, in main model = build_model(args, model_cfg, device, args.num_classes, False, deploy=True) │ │ │ │ │ └ 2 │ │ │ │ └ Namespace(batch_size=1, conf_thresh=0.1, dynamic=True, exp_file=None, experiment_name=None, fuse_conv_bn=False, img_size=640,... │ │ │ └ device(type='cpu') │ │ └ {'backbone': 'elannet', 'pretrained': True, 'bk_act': 'silu', 'bk_norm': 'BN', 'bk_depthwise': True, 'width': 0.25, 'depth': ... │ └ Namespace(batch_size=1, conf_thresh=0.1, dynamic=True, exp_file=None, experiment_name=None, fuse_conv_bn=False, img_size=640,... └ <function build_model at 0x7fe19301d790> File "/home/david/code/paper/PyTorch_YOLO_Tutorial/tools/../models/detectors/__init__.py", line 56, in build_model model, criterion = build_rtcdet( └ <function build_rtcdet at 0x7fe190a9d5e0> File "/home/david/code/paper/PyTorch_YOLO_Tutorial/tools/../models/detectors/rtcdet/build.py", line 26, in build_rtcdet nms_class_agnostic=args.nms_class_agnostic └ Namespace(batch_size=1, conf_thresh=0.1, dynamic=True, exp_file=None, experiment_name=None, fuse_conv_bn=False, img_size=640,... AttributeError: 'Namespace' object has no attribute 'nms_class_agnostic'
@huihui308 Thanks for your reminder, I have fixed this bug just now. You could try again~
I found there is a error if it did not add '--nms_class_agnostic' in export_onnx.py.