Rotation Matters: Generalized Monocular 3D Object Detection for Various Camera Systems
This addresses the generalization issue in 3D object detection for autonomous vehicles, enabling detectors to work across various camera setups without retraining, though it is incremental as it builds on existing networks.
The paper tackles the problem of monocular 3D object detection performance degrading when applied to camera systems different from training datasets, such as from cars to buses, and proposes a compensation module that corrects 3D bounding box location and heading direction, increasing AP3D scores by 6-to-10 times on KITTI moderate without additional training.
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.