Towards Fast, Accurate and Stable 3D Dense Face Alignment

Benchmark Model Rank Results
3d-face-reconstruction-on-aflw2000-3d3DDFA-V2#3Mean NME: 3.56%
3d-face-reconstruction-on-florence3DDFA_V2#10Mean NME: 3.56
3d-face-reconstruction-on-now-benchmark-13DDFA_V2#9Median Reconstruction Error: 1.23Mean Reconstruction Error (mm): 1.57
3d-face-reconstruction-on-realy3DDFA-v2#9all: 1.926@nose: 1.903 (±0.517)@mouth: 1.597 (±0.478)
3d-face-reconstruction-on-realy-side-view3DDFA-v2#6all: 1.943@nose: 1.883 (±0.499)@mouth: 1.642 (±0.501)
3d-face-reconstruction-on-stirling-hq-fg20183DDFA_V2#2Mean Reconstruction Error (mm): 1.91
3d-face-reconstruction-on-stirling-lq-fg20183DDFA_V2#3Mean Reconstruction Error (mm): 2.10
face-alignment-on-aflw2000-3d3DDFA_V2#6Balanced NME (2D Sparse Alignment): 3.51%