Pre One:Efficient Adaptation of Large Vision Transformer via Adapter Re-Composing
Next One:CLIP-guided continual novel class discovery