DBQ-SSD: Dynamic Ball Query for Efficient 3D Object Detection
This work addresses computational inefficiency in 3D object detection for autonomous driving applications, representing an incremental improvement over existing methods.
The paper tackles the problem of inefficient point-based 3D object detection by proposing a Dynamic Ball Query network that adaptively selects points and assigns receptive fields, resulting in a 30%-100% increase in inference speed without performance degradation, reaching up to 162 FPS on KITTI.
Many point-based 3D detectors adopt point-feature sampling strategies to drop some points for efficient inference. These strategies are typically based on fixed and handcrafted rules, making it difficult to handle complicated scenes. Different from them, we propose a Dynamic Ball Query (DBQ) network to adaptively select a subset of input points according to the input features, and assign the feature transform with a suitable receptive field for each selected point. It can be embedded into some state-of-the-art 3D detectors and trained in an end-to-end manner, which significantly reduces the computational cost. Extensive experiments demonstrate that our method can increase the inference speed by 30%-100% on KITTI, Waymo, and ONCE datasets. Specifically, the inference speed of our detector can reach 162 FPS on KITTI scene, and 30 FPS on Waymo and ONCE scenes without performance degradation. Due to skipping the redundant points, some evaluation metrics show significant improvements. Codes will be released at https://github.com/yancie-yjr/DBQ-SSD.