DSFNet: Dual Space Fusion Network for Occlusion-Robust 3D Dense Face Alignment

Input 3DDFA-V2 SynergyNet SADRNet Ours Input 3DDFA-V2 SynergyNet SADRNet Ours Figure 1. The qualitative results compared to state-of-the-art methods on severely occluded or large view angle cases from datasets [38,47]. For each case, from left to right are input, results of 3DDFA-V2 [12], SynergyNet [41], SADRNet [26], and our method.