v1.1_FRoST_class-iNCD Release:

1. Added test-only code for using trained model weights to test the performance directly.
2. Uploaded our trained model weights for reproducing the experimental results in the paper.
3. Solved the bug of making shallow copy instead of deep copy (np.copy() is needed) in the evaluation function due to NumPy version changes.

This release is the PyTorch implementation for the paper Class-incremental Novel Class Discovery (ECCV 2022).