Will you release Pretrained Model based on RGB and ToTensor() Inputs ?

Question

Alexandmrwh opened this issue 5 years ago · 2 comments

Great works! The pretrained model you release is based on bgr input (0-255), will you release the rgb input (transformed by pytorch) ?

Answer 1 · 2019-09-23T07:07:28.000Z

Why don't use the ToTensor() in PyTorch?

use ToBGRTensor() and OpenCVReize() for much faster ? or for better accuracy ?

Answer 2 · 2019-10-09T08:00:36.000Z

Hi @Alexandmrwh , the models we trained are based on BGR input, you may need to adjust your input format if you use the pretrained models.

Hi @zhjpqq , it doesn't affect speed and accuracy, we use it just to keep the data format consistent with our inner framework's format.