Page 3 of 3

Re: YOLO v5 recognition

Posted: Mon 07 Sep 07 2020 3:29 pm
by Daniel Wee
Attempting to retrain with the Windows 10 setup:-

Code: Select all

python train.py --img 640 --batch 16 --epochs 300 --data training2/data.yaml --cfg training2/yolov5l.yaml --weights ''
Results: FAIL

There must be some fundamental problem preventing training on Windows 10. Inference works though.

Re: YOLO v5 recognition

Posted: Mon 14 Sep 14 2020 1:48 pm
by Daniel Wee
We can further speed up inference if TensorRT cores are available (NVIDIA V100-A100 GPUs). This is especially useful for implementations on the TX2 or AGX (or even NX) boards.

https://github.com/wang-xinyu/tensorrtx ... ter/yolov5

Further information here:-

https://www.learnopencv.com/how-to-conv ... inference/