TensorRT10.2.0.19生成.engine报错

[buptxlsk]

Env

• GPU, RTX3050
• OS,Ubuntu22.04
• Cuda 12.2
• TensorRT 10.2.0.19

About this repo

• detect
• YoloV8

Your problem

/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp:109:20: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getNbBindings’
109 | assert(engine->getNbBindings() == 3);
| ^~~~~~~~~~~~~
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp:112:36: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getBindingIndex’
112 | const int inputIndex = engine->getBindingIndex(kInputTensorName);
| ^~~~~~~~~~~~~~~
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp:113:37: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getBindingIndex’
113 | const int outputIndex = engine->getBindingIndex(kOutputTensorName);
| ^~~~~~~~~~~~~~~
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp:114:41: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getBindingIndex’
114 | const int outputIndex_seg = engine->getBindingIndex("proto");
| ^~~~~~~~~~~~~~~
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp: In function ‘void infer(nvinfer1::IExecutionContext&, CUstream_st*&, void**, float*, float*, int, float*, float*, int, std::string)’:
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp:143:13: error: ‘class nvinfer1::IExecutionContext’ has no member named ‘enqueue’; did you mean ‘enqueueV3’?
143 | context.enqueue(batchsize, buffers, stream, nullptr);
| ^~~~~~~
| enqueueV3
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp: In function ‘int main(int, char**)’:
/home/xlsk/Code/tensorrtx/yolov8/yolov8_det.cpp:250:29: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getBindingDimensions’
250 | auto out_dims = engine->getBindingDimensions(1);
| ^~~~~~~~~~~~~~~~~~~~
make[2]: *** [CMakeFiles/yolov8_det.dir/build.make:76：CMakeFiles/yolov8_det.dir/yolov8_det.cpp.o] 错误 1
make[1]: *** [CMakeFiles/Makefile2:117：CMakeFiles/yolov8_det.dir/all] 错误 2
make: *** [Makefile:91：all] 错误 2

[buptxlsk]

请问是API接口更新问题吗？新人不太懂，谢谢。

[mpj1234]

是的不支持，我已经提交了新的yolov8的tensorrt10支持的pr，可以查看 #1557， 或者等待合并到trt10分支

[xiangnameduoganshaya]

是的不支持，我已经提交了新的yolov8的tensorrt10支持的pr，可以查看 #1557， 或者等待合并到trt10分支

你好，我使用了你的代码，生成了可执行文件yolov8_det，但是在构建引擎时出现了这个错误，Loading weights: best.wts
Your platform support int8: true
Building engine, please wait for a while...
reading calib cache: int8calib.table
[09/20/2024-18:23:38] [W] [TRT] Heuristics has been ignored in this builder run. This feature is only supported on Ampere and beyond.
[09/20/2024-18:23:41] [E] [TRT] Unexpected exception _Map_base::at
Build engine successfully!
yolov8_det: /home/xqy/Downloads/tensorrtx-master/yolov8/yolov8_det.cpp:29: void serialize_engine(std::string&, std::string&, int&, std::string&, float&, float&, int&): Assertion `serialized_engine' failed.
Aborted
我不知道是serialize_engine这个函数的问题还是什么问题，希望您能指教一下，十分感谢。

[suojinhui]

engine->getBindingDimensions(1)这个函数已经没了，换成
m_inputDims = m_engine->getTensorShape(m_engine->getIOTensorName(0));
m_outputDims = m_engine->getTensorShape(m_engine->getIOTensorName(1));
类似的写法来获取输入输出的维度

[Q-wenny]

是的不支持，我已经提交了新的yolov8的tensorrt10支持的pr，可以查看 #1557， 或者等待合并到trt10分支

你好，我使用了你的代码，生成了可执行文件yolov8_det，但是在构建引擎时出现了这个错误，Loading weights: best.wts Your platform support int8: true Building engine, please wait for a while... reading calib cache: int8calib.table [09/20/2024-18:23:38] [W] [TRT] Heuristics has been ignored in this builder run. This feature is only supported on Ampere and beyond. [09/20/2024-18:23:41] [E] [TRT] Unexpected exception _Map_base::at Build engine successfully! yolov8_det: /home/xqy/Downloads/tensorrtx-master/yolov8/yolov8_det.cpp:29: void serialize_engine(std::string&, std::string&, int&, std::string&, float&, float&, int&): Assertion `serialized_engine' failed. Aborted 我不知道是serialize_engine这个函数的问题还是什么问题，希望您能指教一下，十分感谢。

你好，我遇到了和你一样的问题，请问你最后解决了吗？

[mpj1234]

是的不支持，我已经提交了新的yolov8的tensorrt10支持的pr，可以查看 #1557， 或者等待合并到trt10分支

你好，我使用了你的代码，生成了可执行文件yolov8_det，但是在构建引擎时出现了这个错误，Loading weights: best.wts Your platform support int8: true Building engine, please wait for a while... reading calib cache: int8calib.table [09/20/2024-18:23:38] [W] [TRT] Heuristics has been ignored in this builder run. This feature is only supported on Ampere and beyond. [09/20/2024-18:23:41] [E] [TRT] Unexpected exception _Map_base::at Build engine successfully! yolov8_det: /home/xqy/Downloads/tensorrtx-master/yolov8/yolov8_det.cpp:29: void serialize_engine(std::string&, std::string&, int&, std::string&, float&, float&, int&): Assertion `serialized_engine' failed. Aborted 我不知道是serialize_engine这个函数的问题还是什么问题，希望您能指教一下，十分感谢。

你的fp16模型能测试通过？我提交代码时候测试int8没问题