Stereo CenterNet based 3D Object Detection for Autonomous Driving

TL;DR

This paper introduces Stereo CenterNet, a stereo image-based 3D object detection method that efficiently predicts 3D bounding boxes using geometric cues, achieving a favorable speed-accuracy balance for autonomous driving.

Contribution

The paper presents a novel stereo image-based 3D detection approach that predicts semantic key points and optimizes bounding boxes with a photometric alignment module, reducing computational costs.

Findings

Achieves state-of-the-art speed-accuracy trade-off on KITTI dataset.

Does not require extra data for training or inference.

Outperforms existing methods in real-time 3D detection accuracy.

Abstract

Recently, three-dimensional (3D) detection based on stereo images has progressed remarkably; however, most advanced methods adopt anchor-based two-dimensional (2D) detection or depth estimation to address this problem. Nevertheless, high computational cost inhibits these methods from achieving real-time performance. In this study, we propose a 3D object detection method, Stereo CenterNet (SC), using geometric information in stereo imagery. SC predicts the four semantic key points of the 3D bounding box of the object in space and utilizes 2D left and right boxes, 3D dimension, orientation, and key points to restore the bounding box of the object in the 3D space. Subsequently, we adopt an improved photometric alignment module to further optimize the position of the 3D bounding box. Experiments conducted on the KITTI dataset indicate that the proposed SC exhibits the best speed-accuracy trade-off among advanced methods without using extra data.