Aiming at the problem that it takes too long to manually label numerous semantic segmentation data sets of vehicle images, a fast and effective data augmentation for semantic segmentation is proposed.

Real-world data is often imbalanced, such that the number of training instances varies by class. Data augmentation (DA) of under-represented classes is commonly used to improve model generalization in ...

I noticed that that in the dataset processing code, a random center crop may be applied to images during training for data augmentation purposes. I was wondering if you noticed this causing any issues ...