3D scene reconstruction on DeepBlending benchmark leaderboard · SOTA2 Research

3D scene reconstruction on DeepBlending

30.24PSNRSizeGS

Evaluation Results

Method	Links
SizeGS2025.05		30.24	90.3	0.271	7.92
Scaffold-GS2025.05		30.21	90.6	0.254	66
SHTC2025.05		30.19	90.3	0.272	2.49
LocoGS2025.05		30.17	90.6	0.244	13.39
HAC++2025.05		30.16	90.7	0.266	2.91
FP-Net2025.05		30.14	90.6	0.274	3.46
Context-GS2025.05		30.11	90.7	0.258	3.45
FP-GS2025.05		30.05	90.6	0.266	3.91
HAC2025.05		29.98	90.2	0.269	4.35
NeuralGS2025.05		29.91	90.6	0.254	12.98
Navaneet et al.2025.05		29.9	90.7	0.251	13.5
CodecGS2025.05		29.82	90.7	0.251	8.62
Compact3DGS2025.05		29.79	90.1	0.258	43.21
EAGLES2025.05		29.72	90.6	0.249	52.34
CompGS2025.05		29.69	90.1	0.279	8.77
Reduced3DGS2025.05		29.63	90.2	0.249	18
RDOGaussian2025.05		29.63	90.2	0.252	18
FCGS2025.05		29.58	89.9	0.248	54.5
MesonGS2025.05		29.51	90.1	0.251	24.76
3DGS2025.05		29.42	89.9	0.247	663.9
Ours30K2026.02		29.41	90.3	0.243	—
Mip-NeRF3602026.02		29.4	90.1	0.245	—
Compressed3D2025.05		29.38	89.8	0.253	25.3
FCGS2025.05		29.27	89.3	0.257	30.1
SOG2025.05		29.12	89.2	0.27	5.7
PUP 3D-GS2025.05		28.85	88.1	0.302	69.92
FlexGaussian2025.05		28.61	88.4	0.269	25.48
Ours7K2026.02		27.78	87.5	0.317	—
LightGaussian2025.05		27.01	87.2	0.308	33.94
Instant-NGP (Base)2026.02		23.62	79.7	0.423	—