Pytorch implementation for "Towards robust classification of multi-view remote sensing images with partial data availability".
This framework not only fully mines the features of multi-view images and improves performance under complete views, but also maintains robustness under missing view without relying on restore and retrieval data.
1.Get the initial weughts for student model and teacher model through train.py;
2.The teacher model and student model learn from each other through train_ml.py.