Detecting As Labeling: Rethinking LiDAR-camera Fusion in 3D Object Detection
This addresses a fundamental issue in 3D object detection for autonomous systems, offering a new baseline that is incremental in method but impactful for practical deployment.
The paper tackles overfitting in LiDAR-camera fusion for 3D object detection by proposing a novel paradigm called DAL, which imitates data annotation to simplify prediction and training, resulting in improved performance and a better speed-accuracy trade-off.
3D object Detection with LiDAR-camera encounters overfitting in algorithm development which is derived from the violation of some fundamental rules. We refer to the data annotation in dataset construction for theory complementing and argue that the regression task prediction should not involve the feature from the camera branch. By following the cutting-edge perspective of 'Detecting As Labeling', we propose a novel paradigm dubbed DAL. With the most classical elementary algorithms, a simple predicting pipeline is constructed by imitating the data annotation process. Then we train it in the simplest way to minimize its dependency and strengthen its portability. Though simple in construction and training, the proposed DAL paradigm not only substantially pushes the performance boundary but also provides a superior trade-off between speed and accuracy among all existing methods. With comprehensive superiority, DAL is an ideal baseline for both future work development and practical deployment. The code has been released to facilitate future work on https://github.com/HuangJunJie2017/BEVDet.