Bobo-y / flexible-yolov5

More readable and flexible yolov5 with more backbone(gcn, resnet, shufflenet, moblienet, efficientnet, hrnet, swin-transformer, etc) and (cbam,dcn and so on), and tensorrt

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

does swin transformer have to do input normalization?

sarmientoj24 opened this issue · comments

commented

@sarmientoj24 I think it can be trained without normalization, but I don't know what resukt it can achieve.

what is the normalization you are using here?

commented

what is the normalization you are using here?

imgs = imgs.to(device, non_blocking=True).float() / 255 # uint8 to float32, 0-255 to 0.0-1.0