Hardness-Aware Scene Synthesis for Semi-Supervised 3D Object Detection

TL;DR

This paper introduces a hardness-aware scene synthesis method that generates diverse, high-quality synthetic scenes to enhance semi-supervised 3D object detection, reducing reliance on costly annotations and improving model generalization.

Contribution

The paper proposes a novel hardness-aware scene synthesis approach that adaptively generates synthetic scenes with pseudo-labels, improving semi-supervised 3D detection performance.

Findings

Outperforms existing semi-supervised methods on KITTI and Waymo datasets.

Effectively reduces impact of low-quality pseudo-labels.

Enhances model generalization through diverse synthetic scene generation.

Abstract

3D object detection aims to recover the 3D information of concerning objects and serves as the fundamental task of autonomous driving perception. Its performance greatly depends on the scale of labeled training data, yet it is costly to obtain high-quality annotations for point cloud data. While conventional methods focus on generating pseudo-labels for unlabeled samples as supplements for training, the structural nature of 3D point cloud data facilitates the composition of objects and backgrounds to synthesize realistic scenes. Motivated by this, we propose a hardness-aware scene synthesis (HASS) method to generate adaptive synthetic scenes to improve the generalization of the detection models. We obtain pseudo-labels for unlabeled objects and generate diverse scenes with different compositions of objects and backgrounds. As the scene synthesis is sensitive to the quality of pseudo-labels, we further propose a hardness-aware strategy to reduce the effect of low-quality pseudo-labels and maintain a dynamic pseudo-database to ensure the diversity and quality of synthetic scenes. Extensive experimental results on the widely used KITTI and Waymo datasets demonstrate the superiority of the proposed HASS method, which outperforms existing semi-supervised learning methods on 3D object detection. Code: https://github.com/wzzheng/HASS.