LiRCDepth: Lightweight Radar-Camera Depth Estimation via Knowledge Distillation and Uncertainty Guidance

TL;DR

LiRCDepth is a lightweight radar-camera depth estimation model that uses knowledge distillation and uncertainty guidance to improve accuracy while maintaining efficiency, achieving notable performance gains on nuScenes dataset.

Contribution

The paper introduces a novel lightweight depth estimation model that incorporates multi-level knowledge distillation and uncertainty guidance, enhancing performance without increasing computational complexity.

Findings

Achieves 6.6% MAE improvement on nuScenes dataset.

Effectively transfers knowledge from complex models to lightweight models.

Demonstrates improved depth estimation accuracy with efficient computation.

Abstract

Recently, radar-camera fusion algorithms have gained significant attention as radar sensors provide geometric information that complements the limitations of cameras. However, most existing radar-camera depth estimation algorithms focus solely on improving performance, often neglecting computational efficiency. To address this gap, we propose LiRCDepth, a lightweight radar-camera depth estimation model. We incorporate knowledge distillation to enhance the training process, transferring critical information from a complex teacher model to our lightweight student model in three key domains. Firstly, low-level and high-level features are transferred by incorporating pixel-wise and pair-wise distillation. Additionally, we introduce an uncertainty-aware inter-depth distillation loss to refine intermediate depth maps during decoding. Leveraging our proposed knowledge distillation scheme, the lightweight model achieves a 6.6% improvement in MAE on the nuScenes dataset compared to the model trained without distillation. Code: https://github.com/harborsarah/LiRCDepth