Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervision

Benchmark Model Rank Results
3d-face-reconstruction-on-now-benchmark-1RingNet#7Median Reconstruction Error: 1.21Mean Reconstruction Error (mm): 1.53
3d-face-reconstruction-on-realyRingNet#16all: 2.258@nose: 1.934 (±0.458)@mouth: 2.074 (±0.616)
3d-face-reconstruction-on-realy-side-viewRingNet#13all: 2.256@nose: 1.921 (±0.451)@mouth: 1.994 (±0.604)
3d-face-reconstruction-on-stirling-hq-fg2018RingNet#3Mean Reconstruction Error (mm): 2.02
3d-face-reconstruction-on-stirling-lq-fg2018RingNet#2Mean Reconstruction Error (mm): 2.08