Reliable Student: Addressing Noise in Semi-Supervised 3D Object Detection

TL;DR

This paper introduces the Reliable Student framework for semi-supervised 3D object detection, effectively mitigating noisy pseudo-label errors through class-aware target assignment and reliability weighting, leading to state-of-the-art results on KITTI.

Contribution

It proposes a novel framework that addresses pseudo-label noise in semi-supervised 3D detection using class-aware strategies and confidence-based reliability weights.

Findings

Achieves 6.2% AP improvement for pedestrians with 1% labeled data.

Surpasses previous state-of-the-art on KITTI benchmark.

Significant gains in AP for pedestrians and cyclists with minimal labeled data.

Abstract

Semi-supervised 3D object detection can benefit from the promising pseudo-labeling technique when labeled data is limited. However, recent approaches have overlooked the impact of noisy pseudo-labels during training, despite efforts to enhance pseudo-label quality through confidence-based filtering. In this paper, we examine the impact of noisy pseudo-labels on IoU-based target assignment and propose the Reliable Student framework, which incorporates two complementary approaches to mitigate errors. First, it involves a class-aware target assignment strategy that reduces false negative assignments in difficult classes. Second, it includes a reliability weighting strategy that suppresses false positive assignment errors while also addressing remaining false negatives from the first step. The reliability weights are determined by querying the teacher network for confidence scores of the student-generated proposals. Our work surpasses the previous state-of-the-art on KITTI 3D object detection benchmark on point clouds in the semi-supervised setting. On 1% labeled data, our approach achieves a 6.2% AP improvement for the pedestrian class, despite having only 37 labeled samples available. The improvements become significant for the 2% setting, achieving 6.0% AP and 5.7% AP improvements for the pedestrian and cyclist classes, respectively.