We present SegLLM, a novel multi-round interactive segmentation model that leverages conversational memory of both visual and textual outputs to reason over previously segmented objects and past ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results