1 code implementation • NeurIPS 2023 • Yanshu Zhang, Shichong Peng, Alireza Moazeni, Ke Li
PAPR effectively learns point cloud positions to represent the correct scene geometry, even when the initialization drastically differs from the target geometry.