Use ncnn low level api instead of eigen for float convolutions

JDAI-CV / dabnn

dabnn is an accelerated binary neural networks inference framework for mobile platform

Other

769 stars 102 forks source link

Use ncnn low level api instead of eigen for float convolutions #66

Open daquexian opened 5 years ago

daquexian commented 5 years ago

Eigen is not the optimal choice for deep learning on mobile. It lacks optimization for convolutions (like winograd) and also not focus on mobile devices.

nihui commented 5 years ago

daquexian is awesome !