C2E: Boosting Ego-Only 3D Object Detection via Multi-Teacher Contrastive Knowledge Distillation

Researchers have introduced a method called C2E that improves 3D object detection for autonomous vehicles by using multi-teacher contrastive knowledge distillation. This technique aims to overcome the limitations of onboard sensors, which often struggle with occlusions and narrow fields of view in complex outdoor environments. By distilling information from multiple perspectives, the model enhances the spatial accuracy of ego-only perception systems without requiring additional hardware.