trt 部署 about tensorrt-for-yolo-series HOT 9 CLOSED

如不改这一块转换的onnx无法转出engine
导出的时候再修改，修改是为了得到只有一个输出的onnx模型

from tensorrt-for-yolo-series.

感谢大佬 马上去试试

from tensorrt-for-yolo-series.

Converting Frontend ==> MIL Ops: 90%|█████████████████████████████████████████████████████████▋ | 749/831 [00:00<00:00, 1859.97 ops/s]
CoreML export failure: The updates tensor should have shape (1, 3, 80, 80, 7). Got (1, 3, 80, 80, 2)

Export complete (4.58s). Visualize with https://github.com/lutzroeder/netron.

自己的数据集2类，导出onnx报了这个错误，查看export也没有啥需要改的吧 pytorch1.9 cuda11.1
麻烦大佬帮忙看看

from tensorrt-for-yolo-series.

我发现转的错误：
CoreML export failure: PyTorch convert function for op 'silu_' not implemented.
这个怎么改呀

from tensorrt-for-yolo-series.

CoreML

如果不需要苹果设备部署的话，不需要关注这一条错误，这倒是导出苹果部署模型出错了

from tensorrt-for-yolo-series.

Converting op 1570 : slice
Adding op '1570' of type slice_by_index
Adding op '1570_begin_0' of type const
Adding op '1570_end_0' of type const
Adding op '1570_end_mask_0' of type const
Converting op y.1_internal_tensor_assign_1 : _internal_op_tensor_inplace_copy
Adding op 'y.1_internal_tensor_assign_1' of type torch_tensor_assign
Adding op 'y.1_internal_tensor_assign_1_begin_0' of type const
Adding op 'y.1_internal_tensor_assign_1_end_0' of type const
Adding op 'y.1_internal_tensor_assign_1_stride_0' of type const
Adding op 'y.1_internal_tensor_assign_1_begin_mask_0' of type const
Adding op 'y.1_internal_tensor_assign_1_end_mask_0' of type const
Adding op 'y.1_internal_tensor_assign_1_squeeze_mask_0' of type const
Converting Frontend ==> MIL Ops: 93%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▊ | 1118/1200 [00:00<00:00, 2597.59 ops/s]
CoreML export failure: The updates tensor should have shape (1, 3, 80, 80, 15). Got (1, 3, 80, 80, 2)

我也遇到这个错误了，请问解决了么

from tensorrt-for-yolo-series.

Converting op 1570 : slice Adding op '1570' of type slice_by_index Adding op '1570_begin_0' of type const Adding op '1570_end_0' of type const Adding op '1570_end_mask_0' of type const Converting op y.1_internal_tensor_assign_1 : _internal_op_tensor_inplace_copy Adding op 'y.1_internal_tensor_assign_1' of type torch_tensor_assign Adding op 'y.1_internal_tensor_assign_1_begin_0' of type const Adding op 'y.1_internal_tensor_assign_1_end_0' of type const Adding op 'y.1_internal_tensor_assign_1_stride_0' of type const Adding op 'y.1_internal_tensor_assign_1_begin_mask_0' of type const Adding op 'y.1_internal_tensor_assign_1_end_mask_0' of type const Adding op 'y.1_internal_tensor_assign_1_squeeze_mask_0' of type const Converting Frontend ==> MIL Ops: 93%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▊ | 1118/1200 [00:00<00:00, 2597.59 ops/s] CoreML export failure: The updates tensor should have shape (1, 3, 80, 80, 15). Got (1, 3, 80, 80, 2)

我也遇到这个错误了，请问解决了么

据我所知这个报错和onnx无关啊

from tensorrt-for-yolo-series.

Converting op 1570 : slice Adding op '1570' of type slice_by_index Adding op '1570_begin_0' of type const Adding op '1570_end_0' of type const Adding op '1570_end_mask_0' of type const Converting op y.1_internal_tensor_assign_1 : _internal_op_tensor_inplace_copy Adding op 'y.1_internal_tensor_assign_1' of type torch_tensor_assign Adding op 'y.1_internal_tensor_assign_1_begin_0' of type const Adding op 'y.1_internal_tensor_assign_1_end_0' of type const Adding op 'y.1_internal_tensor_assign_1_stride_0' of type const Adding op 'y.1_internal_tensor_assign_1_begin_mask_0' of type const Adding op 'y.1_internal_tensor_assign_1_end_mask_0' of type const Adding op 'y.1_internal_tensor_assign_1_squeeze_mask_0' of type const Converting Frontend ==> MIL Ops: 93%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▊ | 1118/1200 [00:00<00:00, 2597.59 ops/s] CoreML export failure: The updates tensor should have shape (1, 3, 80, 80, 15). Got (1, 3, 80, 80, 2)
我也遇到这个错误了，请问解决了么

据我所知这个报错和onnx无关啊

[07/20/2022-15:17:25] [TRT] [I] [MemUsageChange] Init CUDA: CPU +450, GPU +0, now: CPU 477, GPU 5540 (MiB)
[07/20/2022-15:17:25] [TRT] [I] [MemUsageSnapshot] Begin constructing builder kernel library: CPU 477 MiB, GPU 5540 MiB
[07/20/2022-15:17:25] [TRT] [I] [MemUsageSnapshot] End constructing builder kernel library: CPU 631 MiB, GPU 5584 MiB
[07/20/2022-15:17:25] [TRT] [W] onnx2trt_utils.cpp:366: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[07/20/2022-15:17:25] [TRT] [W] onnx2trt_utils.cpp:392: One or more weights outside the range of INT32 was clamped
[07/20/2022-15:17:25] [TRT] [E] [graphShapeAnalyzer.cpp::analyzeShapes::1285] Error Code 4: Miscellaneous (IElementWiseLayer Mul_378: broadcast dimensions must be conformable)
Failed to load ONNX file: /media/yolov7/cfg/deploy/yolov7.onnx
In node 378 (parseGraph): INVALID_NODE: Invalid Node - Mul_378
[graphShapeAnalyzer.cpp::analyzeShapes::1285] Error Code 4: Miscellaneous (IElementWiseLayer Mul_378: broadcast dimensions must be conformable)

没管刚刚那个错误，用转出来的onnx去转tensorrt报了这个错

from tensorrt-for-yolo-series.

Converting op 1570 : slice Adding op '1570' of type slice_by_index Adding op '1570_begin_0' of type const Adding op '1570_end_0' of type const Adding op '1570_end_mask_0' of type const Converting op y.1_internal_tensor_assign_1 : _internal_op_tensor_inplace_copy Adding op 'y.1_internal_tensor_assign_1' of type torch_tensor_assign Adding op 'y.1_internal_tensor_assign_1_begin_0' of type const Adding op 'y.1_internal_tensor_assign_1_end_0' of type const Adding op 'y.1_internal_tensor_assign_1_stride_0' of type const Adding op 'y.1_internal_tensor_assign_1_begin_mask_0' of type const Adding op 'y.1_internal_tensor_assign_1_end_mask_0' of type const Adding op 'y.1_internal_tensor_assign_1_squeeze_mask_0' of type const Converting Frontend ==> MIL Ops: 93%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▊ | 1118/1200 [00:00<00:00, 2597.59 ops/s] CoreML export failure: The updates tensor should have shape (1, 3, 80, 80, 15). Got (1, 3, 80, 80, 2)
我也遇到这个错误了，请问解决了么

据我所知这个报错和onnx无关啊

[07/20/2022-15:17:25] [TRT] [I] [MemUsageChange] Init CUDA: CPU +450, GPU +0, now: CPU 477, GPU 5540 (MiB) [07/20/2022-15:17:25] [TRT] [I] [MemUsageSnapshot] Begin constructing builder kernel library: CPU 477 MiB, GPU 5540 MiB [07/20/2022-15:17:25] [TRT] [I] [MemUsageSnapshot] End constructing builder kernel library: CPU 631 MiB, GPU 5584 MiB [07/20/2022-15:17:25] [TRT] [W] onnx2trt_utils.cpp:366: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32. [07/20/2022-15:17:25] [TRT] [W] onnx2trt_utils.cpp:392: One or more weights outside the range of INT32 was clamped [07/20/2022-15:17:25] [TRT] [E] [graphShapeAnalyzer.cpp::analyzeShapes::1285] Error Code 4: Miscellaneous (IElementWiseLayer Mul_378: broadcast dimensions must be conformable) Failed to load ONNX file: /media/yolov7/cfg/deploy/yolov7.onnx In node 378 (parseGraph): INVALID_NODE: Invalid Node - Mul_378 [graphShapeAnalyzer.cpp::analyzeShapes::1285] Error Code 4: Miscellaneous (IElementWiseLayer Mul_378: broadcast dimensions must be conformable)

没管刚刚那个错误，用转出来的onnx去转tensorrt报了这个错

Pytorch 版本的问题升级到最新版解决

from tensorrt-for-yolo-series.