SparseOcc: Rethinking Sparse Latent Representation for Vision-Based Semantic Occupancy Prediction

arXiv:2404.09502v1 Announce Type: new
Abstract: Vision-based perception for autonomous driving requires an explicit modeling of a 3D space, where 2D latent representations are mapped and subsequent 3D operators are applied. However, operating on dense latent spaces introduces a cubic time and space complexity, which limits scalability in terms of perception range or spatial resolution. Existing approaches compress the dense representation using projections like Bird's Eye View (BEV) or Tri-Perspective View (TPV). Although efficient, these projections result in information loss, especially …

abstract arxiv autonomous autonomous driving complexity cs.cv driving however mapped modeling operators perception prediction representation scalability semantic space spaces terms type vision