Issue with training yolov9-c-dseg

[synthdatagit]

I am trying to train the yolov9-c-dseg (models/segment/yolov9-c-dseg.yaml) model, but i get an error during calculation of loss.

I run the training script with the following command:

python segment/train.py --workers 4 --device 0 --batch 8 --data data/data.yaml --img 640 --cfg models/segment/yolov9-c-dseg.yaml --weights '' --name yolov9-c-dseg --hyp hyp.scratch-high.yaml --no-overlap --epochs 300 --close-mosaic 10

and I get the following error:

Traceback (most recent call last):
File "/mnt/data/yolov9/segment/train.py", line 646, in
main(opt)
File "/mnt/data/yolov9/segment/train.py", line 542, in main
train(opt.hyp, opt, device, callbacks)
File "/mnt/data/yolov9/segment/train.py", line 297, in train
loss, loss_items = compute_loss(pred, targets.to(device), masks=masks.to(device).float())
File "/mnt/data/yolov9/utils/segment/loss_tal.py", line 173, in call
batch_size, _, mask_h, mask_w = proto.shape
AttributeError: 'list' object has no attribute 'shape'

I have tried to use fix described in #528 but I think that was related to another issue.

I have also tried training the gelan-c-seg model with the same command:

python segment/train.py --workers 8 --device 0 --batch 16 --data data/data.yaml --img 640 --cfg models/segment/gelan-c-seg.yaml --weights '' --name gelan-c-seg --hyp hyp.scratch-high.yaml --no-overlap --epochs 300 --close-mosaic 10

and this seems to work just fine.

So what is the issue with the yolov9-c-dseg training?