Single and sparse view 3D reconstruction by learning shape priors

In this paper, we aim to reconstruct free-form 3D models from only one or few silhouettes by learning the priorknowledge of a specific class of objects. Instead of heuristically proposing specific regularities and defining parametricmodels as previous research, our shape prior is learned directly from existing 3D models under a framework based onthe Gaussian Process Latent Variable Model (GPLVM). The major contributions of the paper include: 1) a frameworkfor learning the shape prior of the 3D objects, which requires no heuristic of the object, and can be easily generalized tohandle various categories of 3D objects, and 2) novel probabilistic inference schemes for automatically reconstructing3D shapes from the silhouette(s) in the single view or sparse views. Qualitative and quantitative experimental resultson both synthetic and real data demonstrate the efficacy of our new approach.