This research was presented at (Conference on Computer Vision and Pattern Recognition) and focuses on reconstructing 3D human meshes by leveraging the additional viewpoints provided by mirrors in a single image. Key Paper Details:
The image is part of their Mirrored-Human dataset , which consists of diverse images of people in front of mirrors, specifically annotated to help the model learn from both the direct view and the reflected view. TeenMarvel_Lili-Mesh-Mirror-004.jpg
The researchers developed a method to use mirror reflections to resolve the depth ambiguity inherent in single-view 3D human mesh recovery. This research was presented at (Conference on Computer
Mirrored-Human: Reconstructing 3D Human Pose and Shape from a Single Image with a Mirror. TeenMarvel_Lili-Mesh-Mirror-004.jpg