YOLOv6
YOLOv6 copied to clipboard
meituan
iou_loss dfl_loss都一直为0
Before Asking
-
[X] I have read the README carefully. 我已经仔细阅读了README上的操作指引。
-
[X] I want to train my custom dataset, and I have read the tutorials for training your custom data carefully and organize my dataset correctly; (FYI: We recommand you to apply the config files of xx_finetune.py.) 我想训练自定义数据集，我已经仔细阅读了训练自定义数据的教程，以及按照正确的目录结构存放数据集。（FYI: 我们推荐使用xx_finetune.py等配置文件训练自定义数据集。）
-
[X] I have pulled the latest code of main branch to run again and the problem still existed. 我已经拉取了主分支上最新的代码，重新运行之后，问题仍不能解决。
Search before asking
- [ ] I have searched the YOLOv6 issues and found no similar questions.
Question
img record infomation path is:/media/wang/e28966b9-0a67-4a52-9682-14b0fa37f693/wang/dataset/apple_own_lable/slice_image/slice_cat_own_and_scifresh_coco_imageTwoPart/images/.train2017_cache.json Train: Checking formats of images with 8 process(es): 0 image(s) corrupted: 100%|██████████| 11264/11264 [00:01<00:00, 5893.29it/s] Train: Checking formats of labels with 8 process(es): 0 label(s) found, 11264 label(s) missing, 0 label(s) empty, 0 invalid label files: 100%|██████████| 11264/11264 [00:00<00:00, 25882.86it/s] WARNING: No labels found in /media/wang/e28966b9-0a67-4a52-9682-14b0fa37f693/wang/dataset/apple_own_lable/slice_image/slice_cat_own_and_scifresh_coco_imageTwoPart/images/train2017. Train: Final numbers of valid images: 11264/ labels: 11264. 3.1s for dataset initialization. img record infomation path is:/media/wang/e28966b9-0a67-4a52-9682-14b0fa37f693/wang/dataset/apple_own_lable/slice_image/slice_cat_own_and_scifresh_coco_imageTwoPart/images/.val2017_cache.json Val: Checking formats of images with 8 process(es): 0 image(s) corrupted: 100%|██████████| 2856/2856 [00:00<00:00, 2896.31it/s] Val: Checking formats of labels with 8 process(es): 0 label(s) found, 2856 label(s) missing, 0 label(s) empty, 0 invalid label files: 100%|██████████| 2856/2856 [00:00<00:00, 14740.04it/s] WARNING: No labels found in /media/wang/e28966b9-0a67-4a52-9682-14b0fa37f693/wang/dataset/apple_own_lable/slice_image/slice_cat_own_and_scifresh_coco_imageTwoPart/images/val2017. Val: Final numbers of valid images: 2856/ labels: 2856. 1.8s for dataset initialization. Training start...
Epoch lr iou_loss dfl_loss cls_loss
0/1 0.02 0 0 0.07631: 32%|███▏ | 111/352 [0
显示没有标签，可是该coco标签在mmyolo上是正常的： 07/03 23:27:43 - mmengine - INFO - Epoch(train) [1][ 5/704] lr: 1.8939e-05 eta: 0:27:54 time: 1.1937 data_time: 0.9854 memory: 1337 loss: 4.0770 loss_cls: 3.4109 loss_bbox: 0.6660 07/03 23:27:46 - mmengine - INFO - Epoch(train) [1][ 10/704] lr: 4.2614e-05 eta: 0:20:05 time: 0.8624 data_time: 0.7186 memory: 1095 loss: 3.9894 loss_cls: 3.2906 loss_bbox: 0.6987 07/03 23:27:48 - mmengine - INFO - Epoch(train) [1][ 15/704] lr: 6.6288e-05 eta: 0:17:25 time: 0.7502 data_time: 0.6281 memory: 919 loss: 3.9376 loss_cls: 3.2708 loss_bbox: 0.6668 07/03 23:27:50 - mmengine - INFO - Epoch(train) [1][ 20/704] lr: 8.9962e-05 eta: 0:15:39 time: 0.6770 data_time: 0.5661 memory: 978 loss: 3.8579 loss_cls: 3.1878 loss_bbox: 0.6701 07/03 23:27:55 - mmengine - INFO - Epoch(train) [1][ 25/704] lr: 1.1364e-04 eta: 0:16:13 time: 0.7042 data_time: 0.5974 memory: 1007 loss: 3.7611 loss_cls: 3.0825 loss_bbox: 0.6785 07/03 23:27:57 - mmengine - INFO - Epoch(train) [1][ 30/704] lr: 1.3731e-04 eta: 0:15:24 time: 0.6706 data_time: 0.5679 memory: 958 loss: 3.6729 loss_cls: 2.9957 loss_bbox: 0.6772 07/03 23:27:59 - mmengine - INFO - Epoch(train) [1][ 35/704] lr: 1.6098e-04 eta: 0:14:29 time: 0.6336 data_time: 0.5342 memory: 1007 loss: 3.5962 loss_cls: 2.9100 loss_bbox: 0.6862 希望可以提供一个解决方案（我注意到之前有人提出了类似的问题，但没有给出解决方案）
Additional
No response
您应该是数据配置没有配置对。是不是没有 YOLO 格式的 labels txt 文件？ 您看训练日志，没有发现 labels
0 label(s) found, 11264 label(s) missing
