Transformers for Object Detection in Large Point Clouds
This work addresses a bottleneck in applying transformer models to large-scale point clouds for autonomous driving, offering an incremental improvement with practical compatibility for existing systems.
The paper tackles object detection in large point clouds, such as those from autonomous driving lidar data, by proposing TransLPC, a transformer-based model that modifies the architecture for larger inputs and introduces a query refinement technique, achieving improved detection accuracy on the nuScenes dataset.
We present TransLPC, a novel detection model for large point clouds that is based on a transformer architecture. While object detection with transformers has been an active field of research, it has proved difficult to apply such models to point clouds that span a large area, e.g. those that are common in autonomous driving, with lidar or radar data. TransLPC is able to remedy these issues: The structure of the transformer model is modified to allow for larger input sequence lengths, which are sufficient for large point clouds. Besides this, we propose a novel query refinement technique to improve detection accuracy, while retaining a memory-friendly number of transformer decoder queries. The queries are repositioned between layers, moving them closer to the bounding box they are estimating, in an efficient manner. This simple technique has a significant effect on detection accuracy, which is evaluated on the challenging nuScenes dataset on real-world lidar data. Besides this, the proposed method is compatible with existing transformer-based solutions that require object detection, e.g. for joint multi-object tracking and detection, and enables them to be used in conjunction with large point clouds.