A Point-Based Approach to Efficient LiDAR Multi-Task Perception
This addresses the computational burden for autonomous driving systems by enabling more efficient multi-task perception, though it is incremental as it builds on transformer-based methods.
The paper tackles the inefficiency of multi-task networks in point cloud perception by proposing PAttFormer, a point-based architecture that achieves competitive performance while being 3x smaller and 1.4x faster than existing methods, with improvements of +1.7% in mIou for semantic segmentation and +1.7% in mAP for object detection on the nuScenes benchmark.
Multi-task networks can potentially improve performance and computational efficiency compared to single-task networks, facilitating online deployment. However, current multi-task architectures in point cloud perception combine multiple task-specific point cloud representations, each requiring a separate feature encoder and making the network structures bulky and slow. We propose PAttFormer, an efficient multi-task architecture for joint semantic segmentation and object detection in point clouds that only relies on a point-based representation. The network builds on transformer-based feature encoders using neighborhood attention and grid-pooling and a query-based detection decoder using a novel 3D deformable-attention detection head design. Unlike other LiDAR-based multi-task architectures, our proposed PAttFormer does not require separate feature encoders for multiple task-specific point cloud representations, resulting in a network that is 3x smaller and 1.4x faster while achieving competitive performance on the nuScenes and KITTI benchmarks for autonomous driving perception. Our extensive evaluations show substantial gains from multi-task learning, improving LiDAR semantic segmentation by +1.7% in mIou and 3D object detection by +1.7% in mAP on the nuScenes benchmark compared to the single-task models.