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
1.19k stars 315 forks source link

run engine error #174

Open tuteming opened 2 years ago

tuteming commented 2 years ago

程式有2種編譯的方式:sln(test_dll.exe)及CMakeLists.txt(yolo-trt.exe)均在win10 vs2017 對於yolov4(custum weights)2種方式得到的答案都一樣,ok

我有一yolov4(custum config & weights)的精簡網路,依然以yolov4命名(config & weights) test_dll.exe & yolo-trt.exe 在先產生yolov4-kHALF-batch1.engine後,就會進行偵測,得到的答案都一樣,ok but,在已有engine時,程式會載入engine直接進行偵測,

  1. test_dll.exe出現 Loading TRT Engine... Loading Complete! Assertion failed: get3DTensorVolume(m_Engine->getBindingDimensions(tensor.bindingIndex)) == tensor.volume && "Tensor volumes dont match between cfg and engine file \n", file d:\tools_3\yolo- tensorrt-master\modules\yolo.cpp, line 1290 的錯誤

  2. yolo-trt.exe出現整張影像都是錯誤的偵測框.

請指點一下問題所在,謝謝

tuteming commented 2 years ago

我用的是TensorRT 7

tuteming commented 2 years ago

已解決 yolo.cpp comment line 1139-1144 / if (m_NetworkType == "yolov4")//pan { outputTensor.gridSize = (m_InputH / 32) pow(2, 2-_n_yolo_ind); outputTensor.grid_h = (m_InputH / 32) pow(2, 2-_n_yolo_ind); outputTensor.grid_w = (m_InputW / 32) pow(2, 2-_n_yolo_ind); } */