NeuralSRNF: Neural Square Root Normal Fields for the Statistical Shape Analysis and Generation of Nonrigid 3D and 4D Objects
Read the original on arXiv Computer Vision →NeuralSRNF is a new framework that enables efficient statistical shape analysis and generation of genus‑zero 3D and 4D objects undergoing nonrigid deformations. It replaces costly numerical SRNF inversion with a continuous, resolution‑agnostic neural representation that accurately reconstructs shapes and computes inverse SRNF maps in under 3 s, compared to over 10 min for traditional methods. Experiments on multiple datasets show that NeuralSRNF outperforms existing techniques in accuracy and speed across tasks such as geodesic computation, deformation transfer, statistical summarization, and shape generation.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.