search

Megvii-BaseDetection / YOLOX

YOLOX is a high-performance anchor-free YOLO, exceeding yolov3~v5 with MegEngine, ONNX, TensorRT, ncnn, and OpenVINO supported. Documentation: https://yolox.readthedocs.io/
Apache License 2.0
9.19k stars 2.17k forks source link

Issue regarding torch2trt #1119

Open tulbureandreit opened 2 years ago

tulbureandreit commented 2 years ago

Hello,

I have a Nvidia jetson xavier AGX dev kit

NVIDIA Jetson AGX Xavier [16GB]
   * Jetpack 4.6 [L4T 32.6.1]
   * NV Power Mode: MODE_30W_ALL - Type: 3
   * jetson_stats.service: active
 - Libraries:
   * CUDA: 10.2.300
   * cuDNN: 8.2.1.32
   * TensorRT: 8.0.1.6
   * Visionworks: 1.6.0.501
   * OpenCV: 4.5.1 compiled CUDA: YES
   * VPI: ii libnvvpi1 1.1.15 arm64 NVIDIA Vision Programming Interface library
   * Vulkan: 1.2.70

I try to run tools/trt.py and I get a segmentation fault and no other issue raised helped me thus far.

python3 tools/trt.py -n yolox-s -c yolox_s.pth --workspace 8
Segmentation fault (core dumped)

I want to run DeepSORT+YOLOX

Please help.

FateScript commented 2 years ago

How about trying #1118 ?

tulbureandreit commented 2 years ago

@FateScript solved the problem. I did not have pytorch with cuda support installed...

Moreover, I still have a problem that yolox-s works fine and I can use it with deepstream on a jetson xavier but I cannot use yolox-tiny or nano.

FateScript commented 2 years ago

Tiny and nano version uses depthwise conv, could you double check that if jetson xavier supports such an operation?

andreitulbure commented 2 years ago

@FateScript I will check. Thank you!