SUIT: Learning Significance-guided Information for 3D Temporal Detection
This work addresses a critical problem for autonomous driving and robotics by improving 3D temporal detection with reduced resource usage, though it is incremental as it builds on existing temporal fusion methods.
The paper tackles the challenge of effectively and efficiently using temporal information from sequential LiDAR point clouds for 3D object detection in autonomous driving and robotics, proposing SUIT, which simplifies temporal features into sparse formats to reduce memory and computation costs while achieving state-of-the-art performance on nuScenes and Waymo datasets.
3D object detection from LiDAR point cloud is of critical importance for autonomous driving and robotics. While sequential point cloud has the potential to enhance 3D perception through temporal information, utilizing these temporal features effectively and efficiently remains a challenging problem. Based on the observation that the foreground information is sparsely distributed in LiDAR scenes, we believe sufficient knowledge can be provided by sparse format rather than dense maps. To this end, we propose to learn Significance-gUided Information for 3D Temporal detection (SUIT), which simplifies temporal information as sparse features for information fusion across frames. Specifically, we first introduce a significant sampling mechanism that extracts information-rich yet sparse features based on predicted object centroids. On top of that, we present an explicit geometric transformation learning technique, which learns the object-centric transformations among sparse features across frames. We evaluate our method on large-scale nuScenes and Waymo dataset, where our SUIT not only significantly reduces the memory and computation cost of temporal fusion, but also performs well over the state-of-the-art baselines.