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 316 forks source link

terminate called after throwing an instance of 'std::out_of_range' #121

Open henbucuoshanghai opened 3 years ago

henbucuoshanghai commented 3 years ago

Env

About this repo

Your problem

i use the yolo5s model,get the engine and test it all right.. sudo ./yolov5 -s yolov5s.wts yolov5s.engine s sudo ./yolov5 -d yolov5s.engine ../samples when i use fp32 and fp16,everyting is right!!!

but int8 is wrong. Loading weights: yolov5s.wts Your platform support int8: true Building engine, please wait for a while... reading calib cache: int8calib.table [06/02/2021-16:54:38] [W] [TRT] TensorRT was linked against cuDNN 8.0.5 but loaded cuDNN 8.0.4 terminate called after throwing an instance of 'std::out_of_range' what(): _Map_base::at Aborted

what is wrong?

enazoe commented 3 years ago 
sudo ./yolov5 -s yolov5s.wts yolov5s.engine s
sudo ./yolov5 -d yolov5s.engine ../samples

this is not my project ....

henbucuoshanghai commented 3 years ago

yes..... in wrong project.

by the way ,the yolo 3 and yolo4 in this git are trained under the officifal darnket?

enazoe commented 3 years ago

yes..... in wrong project.

by the way ,the yolo 3 and yolo4 in this git are trained under the officifal darnket?

yes

henbucuoshanghai commented 3 years ago

tks for your help,useful

but yolo34 is in c language,how to change it to tensorrt api?using onnx?or directly using the trt api? where is the related code?

enazoe commented 3 years ago

in my project , I use the trt api build the yolov3 yolov4 model,and parse the darknet weight directly.