You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have solved this problem. The image path in the json files generated by tools/dataset_converters/labelme2coco.py are connected using double "\" which cann't be recognized in Linux. So I changed the double "\" to "/" and now the train.py can run normally.
我在自己数据集fine-tune,但总是报错说找不到图像文件(如标题)
实际图像就在这个路径下,我也尝试将config文件里的coco_train_dataset的data_root改成绝对路径,还是报一样的错误
这是为什么呢?debug也没法进去看到底哪一步出错了,这个runner封装的很高
The text was updated successfully, but these errors were encountered: