enazoe

Results 84 comments of enazoe

batchsize 只能设置为1吗

@jstumpin good idea，and welcome pr :)

Cuda failure: unspecified launch failure in file /yolo-tensorrt/modules/yolo.cpp at line 980

这个不太清楚，没在30系列显卡上做过稳定性测试

YOLOV4_TINY 设置的FP32，预测时间20ms，设置INT8也是20ms。

尺寸是从你cfg里来的，不同实现的耗时可能是不同的，还有就是batch数，有些结果是多batch下的，还有就是平台差异

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

检查下内存是否充足

it is failed for Yolov4 transfer to tensorrt! Please Help answer

@shiguoliang which version of your trt ?

it is failed for Yolov4 transfer to tensorrt! Please Help answer

@shiguoliang use trt7.1 and recommend version

Does it work for win32/x86 on Windows

@PharrellWANG If trt and opencv support x32 ，and everything will be ok

int8预测结果不对

int8是会有精度损失，要用训练集的子集作为校准图片

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

``` sudo ./yolov5 -s yolov5s.wts yolov5s.engine s sudo ./yolov5 -d yolov5s.engine ../samples ``` this is not my project ....

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

> yes..... in wrong project. > > by the way ,the yolo 3 and yolo4 in this git are trained under the officifal darnket? yes