MS3D++: Ensemble of Experts for Multi-Source Unsupervised Domain Adaptation in 3D Object Detection

TL;DR

MS3D++ is a self-training framework that uses ensemble predictions and temporal refinement to improve 3D object detection across different lidar domains, significantly reducing domain gap effects.

Contribution

The paper introduces MS3D++, a novel multi-source unsupervised domain adaptation method that fuses ensemble predictions and refines them temporally for robust 3D detection.

Findings

Achieves state-of-the-art results on Waymo, nuScenes, and Lyft datasets.

Performs comparably to models trained with human labels in BEV evaluation.

Effectively handles different lidar densities and domain shifts.

Abstract

Deploying 3D detectors in unfamiliar domains has been demonstrated to result in a significant 70-90% drop in detection rate due to variations in lidar, geography, or weather from their training dataset. This domain gap leads to missing detections for densely observed objects, misaligned confidence scores, and increased high-confidence false positives, rendering the detector highly unreliable. To address this, we introduce MS3D++, a self-training framework for multi-source unsupervised domain adaptation in 3D object detection. MS3D++ generates high-quality pseudo-labels, allowing 3D detectors to achieve high performance on a range of lidar types, regardless of their density. Our approach effectively fuses predictions of an ensemble of multi-frame pre-trained detectors from different source domains to improve domain generalization. We subsequently refine predictions temporally to ensure temporal consistency in box localization and object classification. Furthermore, we present an in-depth study into the performance and idiosyncrasies of various 3D detector components in a cross-domain context, providing valuable insights for improved cross-domain detector ensembling. Experimental results on Waymo, nuScenes and Lyft demonstrate that detectors trained with MS3D++ pseudo-labels achieve state-of-the-art performance, comparable to training with human-annotated labels in Bird's Eye View (BEV) evaluation for both low and high density lidar. Code is available at https://github.com/darrenjkt/MS3D