2018/07/09 by Swaminathan Gurumurthy, Shubham Agrawal, Gurumurthy, Swaminathan +1 · 1 citation
Computer Science · Engineering · #3D Shape Modeling and Analysis #Advanced Numerical Analysis Techniques #Computer Graphics and Visualization Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Processing and 3D Reconstruction
paper · pdf · doi:10.48550/arxiv.1807.03407
openalex publication_date 2018/07/09 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Semantic shape completion is a challenging problem in 3D computer vision\nwhere the task is to generate a complete 3D shape using a partial 3D shape as\ninput. We propose a learning-based approach to complete incomplete 3D shapes\nthrough generative modeling and latent manifold optimization. Our algorithm\nworks directly on point clouds. We use an autoencoder and a GAN to learn a\ndistribution of embeddings for point clouds of object classes. An input point\ncloud with missing regions is first encoded to a feature vector. The\nrepresentations learnt by the GAN are then used to find the best latent vector\non the manifold using a combined optimization that finds a vector in the\nmanifold of plausible vectors that is close to the original input (both in the\nfeature space and the output space of the decoder). Experiments show that our\nalgorithm is capable of successfully reconstructing point clouds with large\nmissing regions with very high fidelity without having to rely on exemplar\nbased database retrieval.\n