Generative Face Parsing Map Guided 3D Face Reconstruction Under Occluded Scenes

Dapeng Zhao,Yue Qi
2024-12-25
Abstract:Over the past few years, single-view 3D face reconstruction methods can produce beautiful 3D models. Nevertheless,the input of these works is unobstructed <a class="link-external link-http" href="http://faces.We" rel="external noopener nofollow">this http URL</a> describe a system designed to reconstruct convincing face texture in the case of <a class="link-external link-http" href="http://occlusion.Motivated" rel="external noopener nofollow">this http URL</a> by parsing facial features,we propose a complete face parsing map generation method guided by <a class="link-external link-http" href="http://landmarks.We" rel="external noopener nofollow">this http URL</a> estimate the 2D face structure of the reasonable position of the occlusion area,which is used for the construction of 3D <a class="link-external link-http" href="http://texture.An" rel="external noopener nofollow">this http URL</a> excellent anti-occlusion face reconstruction method should ensure the authenticity of the output,including the topological structure between the eyes,nose, and mouth. We extensively tested our method and its components, qualitatively demonstrating the rationality of our estimated facial structure. We conduct extensive experiments on general 3D face reconstruction tasks as concrete examples to demonstrate the method's superior regulation ability over existing methods often break <a class="link-external link-http" href="http://down.We" rel="external noopener nofollow">this http URL</a> further provide numerous quantitative examples showing that our method advances both the quality and the robustness of 3D face reconstruction under occlusion scenes.
Computer Vision and Pattern Recognition
What problem does this paper attempt to address?