Table 4.

Quantitative comparison of model generalization ability in indoor scenes.

Method	SUN RGBD		DIODE indoor		ETH-3D indoor		ScanNet
	δ1↑	REL↓	δ1↑	REL↓	δ1↑	REL↓	δ1↑	REL↓
Lee et al.9	0.448	0.528	0.307	0.380	0.332	0.424	0.267	0.367
Bhat et al.14	0.553	0.334	0.265	0.407	0.348	0.375	0.643	0.208
Ranftl et al.3	0.500	0.355	0.439	0.323	0.465	0.294	0.744	0.190
Proposed	0.568	0.328	0.361	0.352	0.343	0.371	0.695	0.193

The best results are shown in bold, and the second best results are italics.