segment/train_dual.py val loss is all zero.

[zhaochenmeng]

Trained with segment/train_dual.py and found that val/box_loss was all zeros Error after uncommenting line 246 in val_dual：Traceback (most recent call last): File "E:\AI\yolov9\segment\train_dual.py", line 648, in main(opt) File "E:\AI\yolov9\segment\train_dual.py", line 544, in main train(opt.hyp, opt, device, callbacks) File "E:\AI\yolov9\segment\train_dual.py", line 347, in train results, maps, _ = validate.run(data_dict, File "D:\Deeplearning\anaconda3\envs\yolov5\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) File "E:\AI\yolov9\segment\val_dual.py", line 248, in run loss += compute_loss(train_out, targets, masks)[1] # box, obj, cls File "E:\AI\yolov9\utils\segment\loss_tal_dual.py", line 183, in call batch_size,_, mask_h, mask_w = proto.shape ValueError: not enough values to unpack (expected 4, got 3)

[jdiaz97]

multi-gpu or single-gpu?

[zhaochenmeng]

single-gpu

[KNzzzzzzzzzz]

Do you solve the problem?

[zhaochenmeng]

你解决了问题吗？

Not yet

[shahriarahmadf]

Currently the val/box_loss are kept zero for the yolov9 code train_dual.py file. Don't know why though Either we have to find out to how get around or wait for the update.

[babakbch]

any solution to this problem so far?