Repository navigation
[Bug] An error occurs when converting RTMDet-INS to ONNX. #2925
Description
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
11/12 10:11:11 - mmengine - INFO - Start pipeline mmdeploy.apis.pytorch2onnx.torch2onnx in subprocess
11/12 10:11:13 - mmengine - WARNING - Failed to search registry with scope "mmdet" in the "Codebases" registry tree. As a workaround, the current "Codebases" registry in "mmdeploy" is used to build instance. This may cause unexpected failure when running the built modules. Please check whether "mmdet" is a correct scope, or whether the registry is initialized.
11/12 10:11:13 - mmengine - WARNING - Failed to search registry with scope "mmdet" in the "mmdet_tasks" registry tree. As a workaround, the current "mmdet_tasks" registry in "mmdeploy" is used to build instance. This may cause unexpected failure when running the built modules. Please check whether "mmdet" is a correct scope, or whether the registry is initialized.
Loads checkpoint by local backend from path: ./convert/seg_rtmdet_test.pth
Process Process-2:
Traceback (most recent call last):
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\multiprocessing\process.py", line 314, in _bootstrap
self.run()
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\multiprocessing\process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\site-packages\mmdeploy\apis\core\pipeline_manager.py", line 107, in __call__
ret = func(*args, **kwargs)
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\site-packages\mmdeploy\apis\pytorch2onnx.py", line 64, in torch2onnx
data, model_inputs = task_processor.create_input(
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\site-packages\mmdeploy\codebase\mmdet\deploy\object_detection.py", line 192, in create_input
cfg = process_model_config(self.model_cfg, imgs, input_shape)
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\site-packages\mmdeploy\codebase\mmdet\deploy\object_detection.py", line 78, in process_model_config
pipeline = [
File "C:\Users\software\anaconda3\envs\mmdeploy_test\lib\site-packages\mmdeploy\codebase\mmdet\deploy\object_detection.py", line 80, in <listcomp>
if transform.type != 'LoadAnnotations'
AttributeError: 'list' object has no attribute 'type'
11/12 10:11:14 - mmengine - ERROR - C:\Users\software\anaconda3\envs\mmdeploy_test\lib\site-packages\mmdeploy\apis\core\pipeline_manager.py - pop_mp_output - 80 - `mmdeploy.apis.pytorch2onnx.torch2onnx` with Call id: 0 failed. exit.`
The ONNX conversion works fine for the RTMDet model, but issues occur only with RTMDet-INS.
Reproduction
python ..\mmdeploy\tools\deploy.py ../mmdeploy/configs/mmdet/instance-seg/instance-seg_rtmdet-ins_onnxruntime_static-640x640.py ./convert/seg_rtmdet_test_config.py ./convert/seg_rtmdet_test.pth ./convert/test.jpg --work-dir ./convert --device cpu
Environment
11/12 10:11:59 - mmengine - INFO - **********Environmental information**********
11/12 10:12:04 - mmengine - INFO - sys.platform: win32
11/12 10:12:04 - mmengine - INFO - Python: 3.10.19 | packaged by Anaconda, Inc. | (main, Oct 21 2025, 16:41:31) [MSC v.1929 64 bit (AMD64)]
11/12 10:12:04 - mmengine - INFO - CUDA available: True
11/12 10:12:04 - mmengine - INFO - MUSA available: False
11/12 10:12:04 - mmengine - INFO - numpy_random_seed: 2147483648
11/12 10:12:04 - mmengine - INFO - GPU 0,1: NVIDIA GeForce RTX 4090
11/12 10:12:04 - mmengine - INFO - CUDA_HOME: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8
11/12 10:12:04 - mmengine - INFO - NVCC: Cuda compilation tools, release 12.8, V12.8.61
11/12 10:12:04 - mmengine - INFO - MSVC: Microsoft (R) C/C++ 최적화 컴파일러 버전 19.44.35207.1(x64)
11/12 10:12:04 - mmengine - INFO - GCC: n/a
11/12 10:12:04 - mmengine - INFO - PyTorch: 2.3.0
11/12 10:12:04 - mmengine - INFO - PyTorch compiling details: PyTorch built with:
- C++ Version: 201703
- MSVC 192930151
- Intel(R) oneAPI Math Kernel Library Version 2021.4-Product Build 20210904 for Intel(R) 64 architecture applications
- Intel(R) MKL-DNN v3.3.6 (Git Hash 86e6af5974177e513fd3fee58425e1063e7f1361)
- OpenMP 2019
- LAPACK is enabled (usually provided by MKL)
- CPU capability usage: AVX512
- CUDA Runtime 12.1
- NVCC architecture flags: -gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-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;-gencode;arch=compute_90,code=sm_90
- CuDNN 8.8.1 (built against CUDA 12.0)
- Magma 2.5.4
- Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=12.1, CUDNN_VERSION=8.8.1, CXX_COMPILER=C:/cb/pytorch_1000000000000/work/tmp_bin/sccache-cl.exe, CXX_FLAGS=/DWIN32 /D_WINDOWS /GR /EHsc /Zc:__cplusplus /bigobj /FS /utf-8 -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOROCTRACER -DUSE_FBGEMM -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE /wd4624 /wd4068 /wd4067 /wd4267 /wd4661 /wd4717 /wd4244 /wd4804 /wd4273, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=2.3.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=OFF, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=OFF, USE_NNPACK=OFF, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF,
11/12 10:12:04 - mmengine - INFO - TorchVision: 0.18.0
11/12 10:12:04 - mmengine - INFO - OpenCV: 4.12.0
11/12 10:12:04 - mmengine - INFO - MMEngine: 0.10.7
11/12 10:12:04 - mmengine - INFO - MMCV: 2.2.0
11/12 10:12:04 - mmengine - INFO - MMCV Compiler: MSVC 192930148
11/12 10:12:04 - mmengine - INFO - MMCV CUDA Compiler: 12.1
11/12 10:12:04 - mmengine - INFO - MMDeploy: 1.3.1+unknown
11/12 10:12:04 - mmengine - INFO -
11/12 10:12:04 - mmengine - INFO - **********Backend information**********
11/12 10:12:04 - mmengine - INFO - tensorrt: None
11/12 10:12:04 - mmengine - INFO - ONNXRuntime: None
11/12 10:12:04 - mmengine - INFO - ONNXRuntime-gpu: 1.22.0
11/12 10:12:04 - mmengine - INFO - ONNXRuntime custom ops: Available
11/12 10:12:04 - mmengine - INFO - pplnn: None
11/12 10:12:04 - mmengine - INFO - ncnn: None
11/12 10:12:04 - mmengine - INFO - snpe: None
11/12 10:12:04 - mmengine - INFO - openvino: None
11/12 10:12:04 - mmengine - INFO - torchscript: 2.3.0+cu121
11/12 10:12:04 - mmengine - INFO - torchscript custom ops: NotAvailable
11/12 10:12:04 - mmengine - INFO - rknn-toolkit: None
11/12 10:12:04 - mmengine - INFO - rknn-toolkit2: None
11/12 10:12:04 - mmengine - INFO - ascend: None
11/12 10:12:04 - mmengine - INFO - coreml: None
11/12 10:12:04 - mmengine - INFO - tvm: None
11/12 10:12:04 - mmengine - INFO - vacc: None
11/12 10:12:04 - mmengine - INFO -
11/12 10:12:04 - mmengine - INFO - **********Codebase information**********
11/12 10:12:04 - mmengine - INFO - mmdet: 3.3.0
11/12 10:12:04 - mmengine - INFO - mmseg: None
11/12 10:12:04 - mmengine - INFO - mmpretrain: None
11/12 10:12:04 - mmengine - INFO - mmocr: None
11/12 10:12:04 - mmengine - INFO - mmagic: None
11/12 10:12:04 - mmengine - INFO - mmdet3d: None
11/12 10:12:04 - mmengine - INFO - mmpose: None
11/12 10:12:04 - mmengine - INFO - mmrotate: None
11/12 10:12:04 - mmengine - INFO - mmaction: None
11/12 10:12:04 - mmengine - INFO - mmrazor: None
11/12 10:12:04 - mmengine - INFO - mmyolo: None
Activity
Petros626 commentedon Nov 18, 2025
maybe it's fixed here: https://github.com/VBTI-development/onedl-mmdeploy