Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network

Benchmark Model Rank Results
3d-face-reconstruction-on-aflw2000-3dPRN#4Mean NME: 3.9625%
3d-face-reconstruction-on-florencePRN#1Mean NME: 3.7551%
3d-face-reconstruction-on-now-benchmark-1PRNet#12Median Reconstruction Error: 1.50Mean Reconstruction Error (mm): 1.98
3d-face-reconstruction-on-realyPRNet#11all: 2.013@nose: 1.923 (±0.518)@mouth: 1.838 (±0.637)
3d-face-reconstruction-on-realy-side-viewPRNet#9all: 2.032@nose: 1.868 (±0.510)@mouth: 1.856 (±0.607)
3d-face-reconstruction-on-stirling-hq-fg2018PRNet#4Mean Reconstruction Error (mm): 2.06
3d-face-reconstruction-on-stirling-lq-fg2018PRNet#4Mean Reconstruction Error (mm): 2.38
face-alignment-on-aflw2000-3dPRN#9Balanced NME (2D Sparse Alignment): 3.62%