Abstract: Few-Shot Class-Incremental Learning (FSCIL) faces a huge stability-plasticity challenge due to continuously learning knowledge from new classes with a small number of training samples ...
Abstract: The introduction of natural language for vision-language (VL) tracking has been proven to improve performance. However, natural language remains under-explored in existing aerial trackers.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results