XD-RCDepth: Lightweight Radar-Camera Depth Estimation with Explainability-Aligned and Distribution-Aware Distillation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913178443579392 |
|---|---|
| author | Sun, Huawei Wang, Zixu Peng, Xiangyuan Ott, Julius Stettinger, Georg Servadei, Lorenzo Wille, Robert |
| author_facet | Sun, Huawei Wang, Zixu Peng, Xiangyuan Ott, Julius Stettinger, Georg Servadei, Lorenzo Wille, Robert |
| contents | Depth estimation remains central to autonomous driving, and radar-camera fusion offers robustness in adverse conditions by providing complementary geometric cues. In this paper, we present XD-RCDepth, a lightweight architecture that reduces the parameters by 29.7% relative to the state-of-the-art lightweight baseline while maintaining comparable accuracy. To preserve performance under compression and enhance interpretability, we introduce two knowledge-distillation strategies: an explainability-aligned distillation that transfers the teacher's saliency structure to the student, and a depth-distribution distillation that recasts depth regression as soft classification over discretized bins. Together, these components reduce the MAE compared with direct training with 7.97% and deliver competitive accuracy with real-time efficiency on nuScenes and ZJU-4DRadarCam datasets. Code: https://github.com/harborsarah/XD_RCDepth |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_13565 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | XD-RCDepth: Lightweight Radar-Camera Depth Estimation with Explainability-Aligned and Distribution-Aware Distillation Sun, Huawei Wang, Zixu Peng, Xiangyuan Ott, Julius Stettinger, Georg Servadei, Lorenzo Wille, Robert Computer Vision and Pattern Recognition Depth estimation remains central to autonomous driving, and radar-camera fusion offers robustness in adverse conditions by providing complementary geometric cues. In this paper, we present XD-RCDepth, a lightweight architecture that reduces the parameters by 29.7% relative to the state-of-the-art lightweight baseline while maintaining comparable accuracy. To preserve performance under compression and enhance interpretability, we introduce two knowledge-distillation strategies: an explainability-aligned distillation that transfers the teacher's saliency structure to the student, and a depth-distribution distillation that recasts depth regression as soft classification over discretized bins. Together, these components reduce the MAE compared with direct training with 7.97% and deliver competitive accuracy with real-time efficiency on nuScenes and ZJU-4DRadarCam datasets. Code: https://github.com/harborsarah/XD_RCDepth |
| title | XD-RCDepth: Lightweight Radar-Camera Depth Estimation with Explainability-Aligned and Distribution-Aware Distillation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.13565 |