search

Tencent / ncnn

ncnn is a high-performance neural network inference framework optimized for the mobile platform
Other
19.72k stars 4.1k forks source link

YOLACT example is very slow #1903

Open GrushBrush opened 4 years ago

GrushBrush commented 4 years ago

Hello I am trying to run the YOLACT example in realtime using my webcam as an input. However the result is very slow on intel cpu / gpu. Sometimes it gives many false readings as well in a loop.

My questions:

  1. Do you have any tips for faster extraction and processing of the yolact?
  2. Why are we not using Yolact++ ? I know the author, maybe he would be willing to help with that
  3. Why is everything so slow and have so many error on intel gpu / cpu?
  4. We are launching a commercial product based on ncnn, could you refer me to a business development contact? We are using NCNN to transmit to holographic displays.

Thanks, Ethan Schur CEO, HoloPert

HwanWoongLee commented 2 years ago

Still slow.....