Decoupled Pseudo-labeling for Semi-Supervised Monocular 3D Object Detection

arXiv:2403.17387v1 Announce Type: new
Abstract: We delve into pseudo-labeling for semi-supervised monocular 3D object detection (SSM3OD) and discover two primary issues: a misalignment between the prediction quality of 3D and 2D attributes and the tendency of depth supervision derived from pseudo-labels to be noisy, leading to significant optimization conflicts with other reliable forms of supervision. We introduce a novel decoupled pseudo-labeling (DPL) approach for SSM3OD. Our approach features a Decoupled Pseudo-label Generation (DPG) module, designed to efficiently generate pseudo-labels by …

3d object 3d object detection abstract arxiv cs.cv detection labeling labels object optimization prediction quality semi-supervised supervision type