[Bug] no valid mask output from object_detection.py with exported rtmdet-ins

[zylo117]

Checklist

• [ ] I have searched related issues but cannot get the expected help.
• [ ] 2. I have read the FAQ documentation but cannot get the expected help.
• [ ] 3. The bug has not been fixed in the latest version.

Describe the bug

after exporting rtmdet-ins with the latest commit, there is no valid mask but empty array output from object_detection.py. However, the visualized image right after the export can show the correct bbox and masks.

But using Detector class in object_detection.py, bbox and label are correct, but all masks are empty.

Reproduction

run exported tensorrt RTMDet-ins (tiny) Detector class in object_detection.py

Environment

02/07 17:25:11 - mmengine - INFO - 

02/07 17:25:11 - mmengine - INFO - **********Environmental information**********
02/07 17:25:12 - mmengine - INFO - sys.platform: linux
02/07 17:25:12 - mmengine - INFO - Python: 3.8.10 (default, Nov 14 2022, 12:59:47) [GCC 9.4.0]
02/07 17:25:12 - mmengine - INFO - CUDA available: True
02/07 17:25:12 - mmengine - INFO - numpy_random_seed: 2147483648
02/07 17:25:12 - mmengine - INFO - GPU 0,1,2,3: NVIDIA GeForce RTX 2080 Ti
02/07 17:25:12 - mmengine - INFO - CUDA_HOME: /usr/local/cuda-11.7
02/07 17:25:12 - mmengine - INFO - NVCC: Cuda compilation tools, release 11.7, V11.7.99
02/07 17:25:12 - mmengine - INFO - GCC: x86_64-linux-gnu-gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
02/07 17:25:12 - mmengine - INFO - PyTorch: 1.13.0+cu117
02/07 17:25:12 - mmengine - INFO - PyTorch compiling details: PyTorch built with:
  - GCC 9.3
  - C++ Version: 201402
  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
  - Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
  - OpenMP 201511 (a.k.a. OpenMP 4.5)
  - LAPACK is enabled (usually provided by MKL)
  - NNPACK is enabled
  - CPU capability usage: AVX2
  - CUDA Runtime 11.7
  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86
  - CuDNN 8.7  (built against CUDA 11.8)
    - Built with CuDNN 8.5
  - Magma 2.6.1
  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.7, CUDNN_VERSION=8.5.0, CXX_COMPILER=/opt/rh/devtoolset-9/root/usr/bin/c++, CXX_FLAGS= -fabi-version=11 -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wunused-local-typedefs -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Werror=cast-function-type -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.13.0, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, 

02/07 17:25:12 - mmengine - INFO - TorchVision: 0.14.0+cu117
02/07 17:25:12 - mmengine - INFO - OpenCV: 4.5.5
02/07 17:25:12 - mmengine - INFO - MMEngine: 0.4.0
02/07 17:25:12 - mmengine - INFO - MMCV: 2.0.0rc3
02/07 17:25:12 - mmengine - INFO - MMCV Compiler: GCC 9.3
02/07 17:25:12 - mmengine - INFO - MMCV CUDA Compiler: 11.7
02/07 17:25:12 - mmengine - INFO - MMDeploy: 1.0.0rc1+a9f8d8c
02/07 17:25:12 - mmengine - INFO - 

02/07 17:25:12 - mmengine - INFO - **********Backend information**********
02/07 17:25:12 - mmengine - INFO - tensorrt:	8.5.2.2
02/07 17:25:12 - mmengine - INFO - tensorrt custom ops:	Available
02/07 17:25:13 - mmengine - INFO - ONNXRuntime:	1.13.1
02/07 17:25:13 - mmengine - INFO - ONNXRuntime-gpu:	None
02/07 17:25:13 - mmengine - INFO - ONNXRuntime custom ops:	Available
02/07 17:25:13 - mmengine - INFO - pplnn:	None
02/07 17:25:13 - mmengine - INFO - ncnn:	1.0.20230121
02/07 17:25:13 - mmengine - INFO - ncnn custom ops:	Available
02/07 17:25:13 - mmengine - INFO - snpe:	None
02/07 17:25:13 - mmengine - INFO - openvino:	None
02/07 17:25:13 - mmengine - INFO - torchscript:	1.13.0
02/07 17:25:13 - mmengine - INFO - torchscript custom ops:	Available
02/07 17:25:14 - mmengine - INFO - rknn-toolkit:	None
02/07 17:25:14 - mmengine - INFO - rknn-toolkit2:	1.4.0-22dcfef4
02/07 17:25:14 - mmengine - INFO - ascend:	None
02/07 17:25:14 - mmengine - INFO - coreml:	None
02/07 17:25:17 - mmengine - INFO - tvm:	0.9.dev240+g4c47676d1
02/07 17:25:17 - mmengine - INFO - 

02/07 17:25:17 - mmengine - INFO - **********Codebase information**********
02/07 17:25:17 - mmengine - INFO - mmdet:	3.0.0rc5
02/07 17:25:17 - mmengine - INFO - mmseg:	None
02/07 17:25:17 - mmengine - INFO - mmcls:	None
02/07 17:25:17 - mmengine - INFO - mmocr:	None
02/07 17:25:17 - mmengine - INFO - mmedit:	None
02/07 17:25:17 - mmengine - INFO - mmdet3d:	None
02/07 17:25:17 - mmengine - INFO - mmpose:	None
02/07 17:25:17 - mmengine - INFO - mmrotate:	None
02/07 17:25:17 - mmengine - INFO - mmaction:	None

Error traceback

No response

[lvhan028]

What's the converted result? You can find output_tensorrt.jpg under the working directory that is passed to tools/deploy.py.

[zylo117]

converted result is ok, but no valid mask output from Detector of object_detection.py

[soulslicer]

Did you figure this out? I have the same issue

https://github.com/open-mmlab/mmdeploy/issues/2735