自定义数据集出错

[1xdwww]

When I use a custom dataset for training, I get these errors every iteration. Are they related to the above cls_loss: 0.0000, mask_loss: 0.0000, iou_score_loss: 0.0000, num_pos: 0.0000, num_neg: 0.0000? If you can help me solve this issue, I would be extremely grateful！

[GH-Fonic]

Hi, I have the same problem as you, instance segmentationtwo classes one is 0.00 and one is nan, however, semantic segmentation is normal。Did you fix ?

[zyjhxd]

When I use a custom dataset for training, I get these errors every iteration. Are they related to the above cls_loss: 0.0000, mask_loss: 0.0000, iou_score_loss: 0.0000, num_pos: 0.0000, num_neg: 0.0000? If you can help me solve this issue, I would be extremely grateful！

您好我也跟您一样，请问解决了嘛

[zyjhxd]

你好，我遇到了和你一样的问题，实例分割两个类一个是 0.00，一个是 nan，但是语义分割正常。你解决了吗？

请问解决了嘛

[zyjhxd]

Hi, I have the same problem as you, instance segmentationtwo classes one is 0.00 and one is nan, however, semantic segmentation is normal。Did you fix ?

[akaiinc]

+1