Let It Flow: Simultaneous Optimization of 3D Flow and Object Clustering

arXiv:2404.08363v1 Announce Type: new
Abstract: We study the problem of self-supervised 3D scene flow estimation from real large-scale raw point cloud sequences, which is crucial to various tasks like trajectory prediction or instance segmentation. In the absence of ground truth scene flow labels, contemporary approaches concentrate on deducing optimizing flow across sequential pairs of point clouds by incorporating structure based regularization on flow and object rigidity. The rigid objects are estimated by a variety of 3D spatial clustering methods. While …

abstract arxiv cloud clustering cs.cv flow instance labels object optimization prediction raw scale segmentation study tasks trajectory truth type