First-person dynamic spatial reasoning on VSI-Bench and UrbanVideo-Bench (remaining tasks) benchmark leaderboard · SOTA2 Research

First-person dynamic spatial reasoning on VSI-Bench and UrbanVideo-Bench (remaining tasks)

38.16Average AccuracyEvent Cascade Pruning

Evaluation Results

Method	Links
Event Cascade Pruning2026.05		38.16	63.16	69.7	24.5	54.36	34.14	32.21	48.73
Event Cascade Pruning2026.05		37.78	65.79	69.7	24.32	52.7	33.94	31.54	50
Event Cascade Pruning2026.05		37.62	73.68	72.73	23.05	54.36	36.95	30.2	52.12
DEP2026.05		36.88	68.4	69.7	24.28	53.5	36.55	28.9	41.5
DivPrune2026.05		36.77	63.16	69.7	24.05	53.53	36.14	29.53	46.61
PDrop2026.05		36.73	65.79	69.7	23.85	53.53	35.54	30.87	44.49
DTD2026.05		36.55	63.16	69.7	24.17	53.11	35.14	29.53	47.03
VTW2026.05		36.45	65.79	69.7	23.97	51.45	35.14	30.2	42.8
DTD2026.05		36.36	63.16	69.7	24.01	51.45	35.74	29.53	46.61
Original2026.05		36.31	65.79	69.7	23.05	51.43	34.74	29.53	45.23
FastV2026.05		36.28	65.79	69.7	23.47	51.87	34.54	28.86	44.49
DivPrune2026.05		36.16	65.79	69.7	23.53	52.28	35.14	30.87	42.8
BTP2026.05		35.95	65.79	69.7	23.31	51.45	34.14	28.19	44.49
FastV2026.05		35.67	65.79	66.67	22.14	50.21	34.74	30.2	47.46
PDrop2026.05		35.67	65.79	69.7	22.16	50.62	34.74	28.86	45.76
DTD2026.05		35.37	65.79	72.73	22.58	52.28	33.53	29.53	47.88
BTP2026.05		35.09	65.79	69.7	22.2	49.38	34.14	30.2	43.64
DivPrune2026.05		34.4	60.53	66.67	20.33	49.79	35.74	27.52	51.27
VTW2026.05		33.84	68.42	72.73	22.44	49.38	34.74	30.87	38.98
BTP2026.05		33.34	65.79	66.67	19.95	51.87	34.74	30.2	47.88
DEP2026.05		33.28	71.1	72.7	21.18	49	35.14	30.2	36.5
PDrop2026.05		33.12	63.16	66.67	21.5	53.53	32.53	31.54	48.31
FastV2026.05		32.64	57.89	63.64	22.79	47.72	32.73	34.23	49.58
VTW2026.05		30.5	42.11	54.55	12.14	47.72	35.54	30.2	41.95
DEP2026.05		29.95	42.1	51.5	11.11	44.1	37.35	30.9	42.3