RunningLeon

@Aspirinkb @AndPuQing Hi, could you guys try with [mmdeploy](https://github.com/open-mmlab/mmdeploy)? The deployment feature in mmseg would be removed in the future.

@dreamnever Pls. set `dynamic-export=False`. `adaptive pool` cannot be exported to ONNX with dynamic shape.

Same issue. Wonder are there any practical reasons that this OP should not be included in TensorRT?

@kevinch-nv Hi, I'm facing the same problem and K of TopK should be dynamic if we feed dynamic input into a model using TensorRT. And in ONNXRuntime the K is...

@cyning911 Hi, mmdeploy does not support yolact from mmdet based on the [list](https://mmdeploy.readthedocs.io/en/latest/04-supported-codebases/mmdet.html). Maybe you could try to support it by referring to maskrcnn.

@phamdinhdat-ai Hi, could you post here your env info with `python tools/check_env.py` and full error message?

@phamdinhdat-ai it works good on my side. Could you run `python tools/regression_test.py --codebase mmdet3d --backends tensorrt --models PointPillars --device cuda:0` and check if it would succeed?

@yangyangsu29 Not able to reproduce, works fine on my side. ``` python tools/deploy.py \ configs/mmdet3d/voxel-detection/voxel-detection_tensorrt_dynamic-kitti.py \ ../mmdetection3d/configs/pointpillars/hv_pointpillars_secfpn_6x8_160e_kitti-3d-3class.py \ ./checkpoints/mmdet3d/pointpillars/hv_pointpillars_secfpn_6x8_160e_kitti-3d-3class_20220301_150306-37dc2420.pth \ ../mmdetection3d/demo/data/kitti/kitti_000008.bin \ --work-dir work-dirs/mmdet3d/pointpillars/trt \ --device cuda:0 [2022-08-12 09:25:46.018] [mmdeploy]...

@yangyangsu29 Hi, cannot reproduce with docker. ### env `python tools/check_env.py` ``` 2022-08-12 09:23:23,114 - mmdeploy - INFO - 2022-08-12 09:23:23,115 - mmdeploy - INFO - **********Environmental information********** 2022-08-12 09:23:24,699 -...

@yangyangsu29 Hi, could you try with torch==1.8.0, there might be something wrong on torch==1.10.0