stormchasingg

@SandyPanda not yet

@SandyPanda is it the reason? #7

same error talked here [https://github.com/jwyang/faster-rcnn.pytorch/issues/226#issuecomment-402459343](url) but i'm not sure about DA_loss

i just fixed by adding `rpn_loss_cls = torch.unsqueeze(rpn_loss_cls, 0)` `rpn_loss_bbox = torch.unsqueeze(rpn_loss_bbox, 0)` `RCNN_loss_cls = torch.unsqueeze(RCNN_loss_cls, 0)` `RCNN_loss_bbox = torch.unsqueeze(RCNN_loss_bbox, 0)` `DA_img_loss_cls = torch.unsqueeze(DA_img_loss_cls, 0)` `DA_ins_loss_cls = torch.unsqueeze(DA_ins_loss_cls, 0)` `tgt_DA_img_loss_cls...

hi, you can use the python script here [https://github.com/yuhuayc/da-faster-rcnn/pull/18/files](url) and split the trainval.txt into train_s.txt and train_t.txt hope it can help you

lambda=0.1 for loss lambda*alpha=0.01 for back propagation

Images train_s: leftImg8bit/train train_t: leftImg8bit_foggy/train test_t: leftImg8bit_foggy/val 3 classes, 'foggy_beta_0.02', totally 500 images Annotations gtFine/train+val i download dataset from cityscapes official web