List your company

Open-vocabulary video instance segmentation on YTVIS 21 (val)

62.2mAPGLEE
30.06438.40746.7555.093Nov 20, 2025

Evaluation results

MethodLinks
GLEEProcessing mode=all NPs at once2025.1162.2
SAM 32025.1157.4
GLEEProcessing mode=one NP at a time2025.1156.5
SAM 3 Detector + T-by-D2025.1156.5
LLMDet + SAM 3 Tracker2025.1131.3