Wrong input of CANNET

Open CnChengGong opened this issue 10 months ago • 1 comments

https://github.com/mingcv/Bread/blob/e0c50680f38ac9f417ef6a233448f94e68f332f2/train_CAN.py#L95-L99 Should be

    colors = self.model_canet(torch.cat([texture_in, cb_in, cr_in, texture_en, cb_en, cr_en],

Mar 19 '25 02:03 CnChengGong

And https://github.com/mingcv/Bread/blob/e0c50680f38ac9f417ef6a233448f94e68f332f2/train_CAN.py#L106 Should be

    image_out = kornia.color.ycbcr_to_rgb(torch.cat([texture_en, cb, cr], dim=1))

Mar 19 '25 08:03 CnChengGong