Reagan1311 / DABNet

Depth-wise Asymmetric Bottleneck for Real-time Semantic Segmentation (BMVC2019)
https://github.com/Reagan1311/DABNet
MIT License
140 stars 36 forks source link

用的cityscapes官网的数据集训练一直报错误 #34

Closed zhongjiansheng closed 4 years ago

zhongjiansheng commented 4 years ago

Traceback (most recent call last): File "train.py", line 314, in train_model(args) File "train.py", line 217, in train_model lossTr, lr = train(args, trainLoader, model, criteria, optimizer, epoch) File "train.py", line 81, in train loss = criterion(output, labels) File "/home/stu/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(*input, **kwargs) File "/home/stu/user/jszhong/Research/DABNet/utils/loss.py", line 87, in forward index = mask_prob.argsort() RuntimeError: merge_sort: failed to synchronize: device-side assert triggered

Reagan1311 commented 4 years ago

your label map should be 0\~18, not color map, which is 0~255. So you should first convert the label map, you can find the relevant code in cityscapesScripts