yangxue0827 / RotationDetection

This is a tensorflow-based rotation detection benchmark, also called AlphaRotate.
https://rotationdetection.readthedocs.io/
Apache License 2.0
1.08k stars 182 forks source link

During the training, the following error occured #92

Closed ga546 closed 2 years ago

ga546 commented 2 years ago

I use my own dataset image

yangxue0827 commented 2 years ago

Did you add your dataset in label_dict.py?

ga546 commented 2 years ago

Yes I do it,and it always accured at step 1980

yangxue0827 commented 2 years ago

Check if the name of my dataset in your cfgs corresponds to label_dict.py. By the way, please paste the dataset information you added in label_dict.py here.

ga546 commented 2 years ago

@yangxue0827 Hi, I can run the code for now by chang the ADD_BOX_IN_TENSORBOARD in cfgs from true to false.But I don't know the reason.Maybe becase I run it on the server? And I want to know the change I did whether affects the true running of the code。

yangxue0827 commented 2 years ago

This does not solve the problem. Your error message indicates that there is a problem with your data set settings, such as the number of categories, category information, etc. This requires you to paste this information into this.

ga546 commented 2 years ago

This is the dataset information there are 3 categories image

yangxue0827 commented 2 years ago

'back_ground': 0, 'bank_card': 1, 'train_ticket': 2, 'bill_carg': 3

By the way, I recommend you to use RSDet for this dataset, as I have had a similar project experience.

ga546 commented 2 years ago

thanks! I know my errors. I will try it again @yangxue0827