Open-vocabulary Part Segmentation on PARS Ad-hoc Concepts benchmark leaderboard · SOTA2 Research

Open-vocabulary Part Segmentation on PARS Ad-hoc Concepts

57.5gIoUVLPart*

Evaluation Results

Method	Links
VLPart*2026.05		57.5	63.2	23.9
VASA2026.05		56.9	66.5	22.9
VLPart*2026.05		55.2	53.8	35.1
VASA2026.05		54	56.9	33.5
Seg-Zero2026.05		47.8	50.8	47.4
SAM3 Agent2026.05		46.4	48.1	48
Semantic-SAM†2026.05		44.5	39.4	38.4
LISA2026.05		42.6	49.4	32.9
LISA++2026.05		42.3	41.5	41
Seg-Zero2026.05		40.8	39.6	58.2
SAM3 Agent2026.05		40.5	39.1	58.8
SegAgent-SC2026.05		40.4	47.9	33.8
OVSeg2026.05		39.3	34.2	33.5
Semantic-SAM†2026.05		39.2	30.4	45.2
SegAgent-SAM2026.05		38.5	43.1	35
RESAnything2026.05		38.2	41.8	38.1
LISA2026.05		37.9	39.9	45
SAM32026.05		37.3	45	48.9
CoReS2026.05		36.9	44	27.5
LISA++2026.05		35.2	33.7	52.6
SegAgent-SC2026.05		34.6	34.8	46.6
OVSeg2026.05		34.4	27.6	42.2
SAM32026.05		34.1	36.1	59.8
RESAnything2026.05		33.1	34	49.7
SegAgent-SAM2026.05		33.1	31.7	48.2
CoReS2026.05		33	34.1	36.2
VisionReasoner2026.05		28.1	35	34.5
VisionReasoner2026.05		27.2	28.8	47.9