Multi-Frame to Single-Frame: Knowledge Distillation for 3D Object Detection
This work addresses a practical challenge in autonomous driving by enabling more robust 3D object detection with sparse inputs, though it is incremental as it builds on existing knowledge distillation techniques.
The paper tackles the problem of 3D object detection in autonomous driving where high-quality point clouds are available only during training, not testing, by using knowledge distillation to transfer knowledge from a model trained on dense multi-frame data to one tested on sparse single-frame data, resulting in improved performance on low-quality data without extra overhead.
A common dilemma in 3D object detection for autonomous driving is that high-quality, dense point clouds are only available during training, but not testing. We use knowledge distillation to bridge the gap between a model trained on high-quality inputs at training time and another tested on low-quality inputs at inference time. In particular, we design a two-stage training pipeline for point cloud object detection. First, we train an object detection model on dense point clouds, which are generated from multiple frames using extra information only available at training time. Then, we train the model's identical counterpart on sparse single-frame point clouds with consistency regularization on features from both models. We show that this procedure improves performance on low-quality data during testing, without additional overhead.