Reconstructing 3D Model from Single-View Sketch with Deep Neural Network

Table 2

Man-made objects (synthetic).

	Ours	ShapeMVD	Nearest retrieval	Tatarchenko et al. [15]	[15]+U-net	Volumetric decoder	R2N2
Hausdorff distance	0.076	0.092	0.165	0.142	0.121	0.113	0.144
Chamfer distance	0.011	0.015	0.025	0.022	0.017	0.021	0.026
Normal distance	26.45	30.66	42.57	35.58	32.32	49.40	48.78
Depth map error	0.013	0.026	0.049	0.039	0.030	0.038	0.045
Volumetric distance	0.276	0.344	0.501	0.442	0.374	0.432	0.512