KDTwin: Task-Aware Knowledge Distillation for Lightweight Multi-Task Driving Scene Segmentation

arXiv:2609.18955v1 Announce Type: new Abstract: Efficient perception models are essential for real-time autonomous driving, where accuracy and computational cost must be carefully balanced. However, applying knowledge distillation to multi-task driving scene segmentation is challenging because drivable-area and lane segmentation exhibit different spatial characteristics and class imbalance. We propose KDTwin, a task-aware distillation framework for lightweight multi-task segmentation networks. The proposed method performs distillation at both the shared encoder and task-specific decoders. Encoder-level pairwise distillation transfers spatial relational knowledge to enhance the student's shared representation. For the decoders, we use a weighted loss for drivable-area segmentation and a boundary-aware loss for lane segmentation, enabling task-adaptive knowledge transfer without increasing inference complexity. Experiments on BDD100K show consistent improvements across the evaluated CNN-based and Transformer-based student models without increasing inference-time parameters or FLOPs. The results show that designing distillation objectives according to task-specific characteristics can effectively enhance multi-task segmentation performance for autonomous driving. The source code is available at https://github.com/chequanghuy/KDTwin.

Related

• Distill, Diffuse, and Semanticize (DDS): Annotation-Free 3D Scene Understanding Based on Multi-Granularity Distillation and Graph-Diffusion-Based Segmentation
• LIE: LiDAR-only HD Map Construction with Intensity Enhancement via Online Knowledge Distillation
• XD-RCDepth: Lightweight Radar-Camera Depth Estimation with Explainability-Aligned and Distribution-Aware Distillation

Source: arXiv cs.CV | 2026-09-17