Luojianlong
Luojianlong
i have the same problem when training the 1% and 2% labeling ratios.Have you solved this problem or we can intercourse for this.
the program can not find out the 10% labeling ratio in fold 1.you need to create the dataset by /tools/dataset/prepare_coco_data.sh.
Have you run into the problem for valueError in training which make me annoyed?
前5000次是预热，只进行全监督训练，5000次后才进行半监督
1.Change the data image path in configs. 2.Change rcnn head num_classes. 3.Change coco.py CLASSES
应该是显存不够吧，模型快20g，还有数据也会占很多
 图片路径是没有问题的，num_worker是指workers_per_gpu的值吗，如果是我就没有设错。但问题是只要运行就会报[ERROR][ModelInfer] Found no image in /home/hello/PycharmProjects/pythonProject/new_DSL/DSL/mydata/semicoco/images/full，我昨天也遇到这个问题了，然后重新再配了一遍代码就没事了，但报错后再运行就会报这个错，所以现在有点困惑。
我把train，unlabel_train/pred，val的图片路径都在.sh设好了，就是有点不清楚这个报错是哪个部分出来的