Fetching the paper…

Large-scale Pre-trained Models are Surprisingly Strong in Incremental Novel Class Discovery · Around