Text-to-Scene Generation on 3D-FRONT Livingroom (test) benchmark leaderboard · SOTA2 Research

Text-to-Scene Generation on 3D-FRONT Livingroom (test)

105.01FIDAccioScene

Evaluation Results

Method	Links
AccioScene2025.02		105.01	5.24	6,780	65.72	57.24
MG-DiT2025.06		108.22	5.23	3.87	54.05	71.81
SceneNAT2026.01		110.28	5.49	0.0062	—	59.56
InstructScene2025.02		110.39	5.37	8,160	65.42	56.81
MG-DiT2025.06		110.55	5.83	7.95	55.24	58.16
FreeScene2026.01		110.55	5.83	0.008	—	58.16
FreeScene2025.02		110.55	5.83	7,950	55.24	58.16
InstructScene2025.06		111.52	5.91	8.65	55.32	57.21
InstructScene2026.01		111.52	5.91	0.0087	—	57.21
ATISS2025.02		117.67	6.08	17,600	69.38	29.5
ATISS2025.06		120.1	7.43	16.44	57.02	36.31
ATISS2026.01		120.1	7.43	0.0164	—	36.31
DiffuScene2025.02		122.2	6.1	16,490	72.92	31.15
DiffuScene2025.06		135.93	10.71	29.07	58.98	39.58
DiffuScene2026.01		135.93	10.71	0.0291	—	39.58