FedS2R: One-Shot Federated Domain Generalization for Synthetic-to-Real Semantic Segmentation in Autonomous Driving

2025-07-26

AI Summary

This work addresses the semantic segmentation generalization gap from synthetic to real-world data in autonomous driving, proposing FedS2R—the first single-round federated domain generalization framework that enables multi-client collaborative training without sharing raw data. Methodologically, it innovatively integrates inconsistency-driven data augmentation, cross-client knowledge distillation, and feature-level fusion to jointly optimize privacy preservation and cross-domain transferability. Extensive experiments across five real-world benchmarks demonstrate that the global model achieves a significantly higher mIoU than all local models, lagging behind centralized joint training by only 2 percentage points—validating its efficacy for efficient cross-domain generalization under federated settings. To the best of our knowledge, this is the first study to systematically apply federated domain generalization to autonomous-driving semantic segmentation, establishing a novel paradigm for robust, privacy-sensitive visual understanding.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Adversarial Attacks & Robustness
• Application Domains: Mobility, Driving & Flight

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Federated domain generalization has shown promising progress in image classification by enabling collaborative training across multiple clients without sharing raw data. However, its potential in the semantic segmentation of autonomous driving remains underexplored. In this paper, we propose FedS2R, the first one-shot federated domain generalization framework for synthetic-to-real semantic segmentation in autonomous driving. FedS2R comprises two components: an inconsistency-driven data augmentation strategy that generates images for unstable classes, and a multi-client knowledge distillation scheme with feature fusion that distills a global model from multiple client models. Experiments on five real-world datasets, Cityscapes, BDD100K, Mapillary, IDD, and ACDC, show that the global model significantly outperforms individual client models and is only 2 mIoU points behind the model trained with simultaneous access to all client data. These results demonstrate the effectiveness of FedS2R in synthetic-to-real semantic segmentation for autonomous driving under federated learning

Problem

• Enabling synthetic-to-real semantic segmentation in autonomous driving
• Addressing federated domain generalization for segmentation tasks
• Improving model performance without sharing raw client data

Innovation

• One-shot federated domain generalization framework
• Inconsistency-driven data augmentation for unstable classes
• Multi-client knowledge distillation with feature fusion