Open KingWangJL opened 5 years ago
Can you show me your time cost by video? It should be approximate.
My model is applied to the nano, we hope to detect hand in real time. And The framework I used is pytorch.
Do you have tried the tensorRT version besides pytorch?
yes,i‘m trying the version of TensorFlow.
You can try the tiny-yolov3 and runing it by TensorRT.
yes,i considered the tiny-yolov3,but the high accuracy is required.
I have completed the official example, but when the input of the model is video, the effect is very poor.i need your help ,thanks.