search

enazoe / yolo-tensorrt

TensorRT8.Support Yolov5n,s,m,l,x .darknet -> tensorrt. Yolov4 Yolov3 use raw darknet *.weights and *.cfg fils. If the wrapper is useful to you,please Star it.
MIT License
1.18k stars 316 forks source link

Compiling Error TensorRT 6.5.0.1 #146

Open hakan6710 opened 2 years ago

hakan6710 commented 2 years ago

Hi,

first of all, thank you for this repo. It works very well on my Laptop.

I am trying to get it working on an NVIDIA Drive AGX with Drive OS 5.2.6. It has TensorRT 6.5.0.1 for ARM64

I have a Problem with compiling the code: :2: /home/nvidia/dev_ws_galactic/src/ros2-node/src/mot/detection/./yolo/modules/detect.h:121:15: error: looser throw specifier for ‘virtual const char* nvinfer1::DetectPluginCreator::getPluginNamespace() const’ const char* getPluginNamespace() const override; ^~~~~~~~~~~~~~~~~~ In file included from /usr/include/aarch64-linux-gnu/NvInferRuntime.h:59:0, from /usr/include/aarch64-linux-gnu/NvInfer.h:53, from /home/nvidia/dev_ws_galactic/src/ros2-node/src/mot/detection/./yolo/modules/mish.h:6, from /home/nvidia/dev_ws_galactic/src/ros2-node/src/mot/detection/./yolo/modules/trt_utils.h:38, from /home/nvidia/dev_ws_galactic/src/ros2-node/src/mot/detection/./yolo/modules/ds_image.h:28, from /home/nvidia/dev_ws_galactic/src/ros2-node/src/mot/detection/./yolo/modules/class_yolo_detector.hpp:5, from /home/nvidia/dev_ws_galactic/src/ros2-node/src/mot/detection/class_detector.cpp:2: /usr/include/aarch64-linux-gnu/NvInferRuntimeCommon.h:904:30: error: overriding ‘virtual const AsciiChar* nvinfer1::IPluginCreator::getPluginNamespace() const noexcept’ virtual AsciiChar const* getPluginNamespace() const noexcept = 0; ^~~~~~~~~~~~~~~~~~ make[2]: *** [CMakeFiles/detector.dir/src/mot/detection/yolo/modules/calibrator.cpp.o] Error 1 make[2]: *** [CMakeFiles/detector.dir/src/mot/detection/yolo/modules/ds_image.cpp.o] Error 1 make[2]: *** [CMakeFiles/detector.dir/src/mot/detection/class_detector.cpp.o] Error 1 make[1]: *** [CMakeFiles/detector.dir/all] Error 2 make: *** [all] Error 2

I am guessing that the API between TensorRT 7.X and 6.X is different. Did you ever test your code with TensorRT 6.X? Do you think that it could work, if I just change the overwritten functions ? I have no knowledge where the differences are between 6.x and 7.x

Thank you in advance for your reply.

enazoe commented 2 years ago

does not support trt 6.x now

hakan6710 commented 2 years ago

Do you think that it would be easy to adapt to it ?

hakan6710 commented 2 years ago

I was able to solve the issue by using the changes from the TensorRT 8 branch. I am guessing that nvidia continues to update tensorRT for the drive platform and 6.5.3 is similiar to 8.X

riziamm commented 1 year ago

I was able to solve the issue by using the changes from the TensorRT 8 branch. I am guessing that nvidia continues to update tensorRT for the drive platform and 6.5.3 is similiar to 8.X

Hi, Could you elaborate on what changes you made from TensorRT 8 branch for it to work with 6.5.3, please?

h-koese commented 1 year ago

I am not sure anymore. I think I used the master branch and changed a few function prototypes. So the changes in "Tensor8" branch were used.

For what do you want to use this repo? Are you working this Drive OS and AGX?

riziamm commented 1 year ago

Hey,

I understand. I want to use it in Jetson Nano and Xavier with TRT 6 for my custom model. Let me know if you can remember some of the changes.

Thank you!

On Mon, Oct 17, 2022, 08:14 Hakan Köse @.***> wrote:

I am not sure anymore. I think I used the master branch and changed a few function prototypes.

For what do you want to use this repo? Are you working this Drive OS and AGX?

— Reply to this email directly, view it on GitHub https://github.com/enazoe/yolo-tensorrt/issues/146#issuecomment-1280929484, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALNGWRI7YW7TSO2SKZY275TWDVNK5ANCNFSM5E5HE57Q . You are receiving this because you commented.Message ID: @.***>

hakan6710 commented 1 year ago

Do you have an Error Message ? Could you show me it ? @RizM9292