DPO: Dual-Perturbation Optimization for Test-time Adaptation in 3D Object Detection

arXiv:2406.13891v1 Announce Type: new
Abstract: LiDAR-based 3D object detection has seen impressive advances in recent times. However, deploying trained 3D detectors in the real world often yields unsatisfactory performance when the distribution of the test data significantly deviates from the training data due to different weather conditions, object sizes, \textit{etc}. A key factor in this performance degradation is the diminished generalizability of pre-trained models, which creates a sharp loss landscape during training. Such sharpness, when encountered during testing, can precipitate …

3d object 3d object detection abstract advances arxiv cs.ai cs.cv data deploying detection detectors distribution dpo etc however lidar object optimization performance test test data training training data type weather world