| RingNet | 2.028 (±0.720) | 2.995 (±0.908) | 2.074 (±0.616) | 1.934 (±0.458) | 2.258 | Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervision | |
| N-3DMM | 1.918 (±0.801) | 4.582 (±1.488) | 2.375 (±0.599) | 2.936 (±0.810) | 2.953 | Towards High-fidelity Nonlinear 3D Face Morphable Model | - |
| PRNet | 1.863 (±0.698) | 2.429 (±0.588) | 1.838 (±0.637) | 1.923 (±0.518) | 2.013 | Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network | |
| SADRNet | 1.856 (±0.701) | 2.413 (±0.537) | 1.591 (±0.488) | 1.791 (±0.542) | 1.913 | SADRNet: Self-Aligned Dual Face Regression Networks for Robust 3D Dense Face Alignment and Reconstruction | |
| 3DDFA-v2 | 1.757 (±0.642) | 2.447 (±0.647) | 1.597 (±0.478) | 1.903 (±0.517) | 1.926 | Towards Fast, Accurate and Stable 3D Dense Face Alignment | |
| ExpNet | 1.717 (±0.590) | 3.084 (±1.005) | 1.912 (±0.450) | 2.509 (±0.486) | 2.306 | ExpNet: Landmark-Free, Deep, 3D Facial Expressions | |
| MGCNet | 1.665 (±0.644) | 2.248 (±0.508) | 1.409 (±0.418) | 1.827 (±0.383) | 1.787 | Self-Supervised Monocular 3D Face Reconstruction by Occlusion-Aware Multi-view Geometry Consistency | |
| SynergyNet | 1.647 (±0.622) | 2.679 (±0.741) | 1.731 (±0.502) | 2.026 (±0.532) | 2.021 | Synergy between 3DMM and 3D Landmarks for Accurate 3D Facial Geometry | |
| PSL | 1.593 (±0.540) | 2.350 (±0.551) | 1.876 (±0.563) | 1.708 (±0.349) | 1.882 | A Perceptual Shape Loss for Monocular 3D Face Reconstruction | - |
| Deep3D | 1.528 (±0.501) | 2.015 (±0.449) | 1.368 (±0.439) | 1.719 (±0.354) | 1.657 | Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set | |
| EMOCA-f | 1.495 (±0.469) | 2.595 (±0.631) | 2.929 (±1.106) | 2.532 (±0.539) | 2.388 | EMOCA: Emotion Driven Monocular Face Capture and Animation | |
| DECA-c | 1.479 (±0.535) | 2.394 (±0.576) | 2.516 (±0.839) | 1.697 (±0.355) | 2.010 | Learning an Animatable Detailed 3D Face Model from In-The-Wild Images | |
| CEST | 1.456 (±0.485) | 2.384 (±0.578) | 1.448 (±0.406) | 2.779 (±0.835) | 2.017 | Self-Supervised 3D Face Reconstruction via Conditional Estimation | - |
| DECA-f | 1.443 (±0.498) | 2.457 (±0.559) | 2.802 (±0.868) | 2.138 (±0.461) | 2.210 | Learning an Animatable Detailed 3D Face Model from In-The-Wild Images | |
| EMOCA-c | 1.438 (±0.501) | 2.426 (±0.641) | 2.679 (±1.112) | 1.868 (±0.387) | 2.103 | EMOCA: Emotion Driven Monocular Face Capture and Animation | |
| HiFace-c | 1.342 (±0.384) | 1.331 (±0.347) | 1.461 (±0.430) | 1.054 (±0.317) | 1.297 | HiFace: High-Fidelity 3D Face Reconstruction by Learning Static and Dynamic Details | - |
| GANFit | 1.329 (±0.504) | 2.402 (±0.545) | 1.812 (±0.544) | 1.928 (±0.490) | 1.868 | GANFIT: Generative Adversarial Network Fitting for High Fidelity 3D Face Reconstruction | |
| HiFace-f | 1.291 (±0.362) | 1.324 (±0.334) | 1.450 (±0.413) | 1.036 (±0.280) | 1.275 | HiFace: High-Fidelity 3D Face Reconstruction by Learning Static and Dynamic Details | - |
| AlbedoGAN | 1.141 (±0.303) | 2.102 (±0.512) | 2.087 (±0.839) | 1.656 (±0.374) | 1.746 | Towards Realistic Generative 3D Face Models | |
| MoSAR | 1.128 (±0.303) | 1.950 (±0.559) | 1.424 (±0.462) | 1.499 (±0.366) | 1.500 | MoSAR: Monocular Semi-Supervised Model for Avatar Reconstruction using Differentiable Shading | - |