Canonical Pose Reconstruction from Single Depth Image for 3D Non-rigid Pose Recovery on Limited Datasets
This addresses the problem of limited training data for 3D non-rigid pose recovery, offering a solution for applications like human and animal modeling, though it appears incremental by adapting existing rigid methods to non-rigid cases.
The study tackled 3D reconstruction of non-rigid objects from single-view depth images by proposing a canonical pose reconstruction model that transforms inputs into a canonical form, enabling the use of rigid object techniques and achieving effective results with only about 300 samples, outperforming state-of-the-art methods on animal and human datasets.
3D reconstruction from 2D inputs, especially for non-rigid objects like humans, presents unique challenges due to the significant range of possible deformations. Traditional methods often struggle with non-rigid shapes, which require extensive training data to cover the entire deformation space. This study addresses these limitations by proposing a canonical pose reconstruction model that transforms single-view depth images of deformable shapes into a canonical form. This alignment facilitates shape reconstruction by enabling the application of rigid object reconstruction techniques, and supports recovering the input pose in voxel representation as part of the reconstruction task, utilizing both the original and deformed depth images. Notably, our model achieves effective results with only a small dataset of approximately 300 samples. Experimental results on animal and human datasets demonstrate that our model outperforms other state-of-the-art methods.