CaKDP: Category-aware Knowledge Distillation and Pruning Framework for Lightweight 3D Object Detection

Haonan Zhang, Longjun Liu, Yuqi Huang, Zhao Yang, Xinyu Lei, Bihan Wen; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 15331-15341

Knowledge distillation (KD) possesses immense potential to accelerate the deep neural networks (DNNs) for LiDAR-based 3D detection. However in most of prevailing approaches the suboptimal teacher models and insufficient student architecture investigations limit the performance gains. To address these issues we propose a simple yet effective Category-aware Knowledge Distillation and Pruning (CaKDP) framework for compressing 3D detectors. Firstly CaKDP transfers the knowledge of two-stage detector to one-stage student one mitigating the impact of inadequate teacher models. To bridge the gap between the heterogeneous detectors we investigate their differences and then introduce the student-motivated category-aware KD to align the category prediction between distillation pairs. Secondly we propose a category-aware pruning scheme to obtain the customizable architecture of compact student model. The method calculates the category prediction gap before and after removing each filter to evaluate the importance of filters and retains the important filters. Finally to further improve the student performance a modified IOU-aware refinement module with negligible computations is leveraged to remove the redundant false positive predictions. Experiments demonstrate that CaKDP achieves the compact detector with high performance. For example on WOD CaKDP accelerates CenterPoint by half while boosting L2 mAPH by 1.61%. The code is available at https://github.com/zhnxjtu/CaKDP.