Rotation Matters: Generalized Monocular 3D Object Detection for Various Camera Systems

TL;DR

This paper identifies how camera orientation affects monocular 3D detection performance and introduces a universal compensation module that significantly improves accuracy across various camera systems without retraining.

Contribution

The paper proposes a generalized 3D detection method with a novel compensation module that corrects for camera pose variations, enhancing cross-system applicability.

Findings

Camera pose, especially orientation, impacts detection accuracy.

The proposed module improves AP3D scores by 6-10 times.

The method works across different camera setups without retraining.

Abstract

Research on monocular 3D object detection is being actively studied, and as a result, performance has been steadily improving. However, 3D object detection performance is significantly reduced when applied to a camera system different from the system used to capture the training datasets. For example, a 3D detector trained on datasets from a passenger car mostly fails to regress accurate 3D bounding boxes for a camera mounted on a bus. In this paper, we conduct extensive experiments to analyze the factors that cause performance degradation. We find that changing the camera pose, especially camera orientation, relative to the road plane caused performance degradation. In addition, we propose a generalized 3D object detection method that can be universally applied to various camera systems. We newly design a compensation module that corrects the estimated 3D bounding box location and heading direction. The proposed module can be applied to most of the recent 3D object detection networks. It increases AP3D score (KITTI moderate, IoU $> 70\%$) about 6-to-10-times above the baselines without additional training. Both quantitative and qualitative results show the effectiveness of the proposed method.