analiza aiqbench
This paper introduces FedS2R, a novel framework for one-shot federated domain generalization specifically targeting synthetic-to-real semantic segmentation in autonomous driving. By combining inconsistency-driven data augmentation with multi-client knowledge distillation, the method effectively bridges the gap between simulated and real-world environments without sharing raw data across clients. The results demonstrate that this collaborative approach yields global models nearly on par with those trained using centralized access to all datasets, marking a significant step forward for privacy-preserving AI in vehicle perception systems.
FedS2R: One-Shot Federated Domain Generalization for Synthetic-to-Real Semantic Segmentation in Autonomous Driving
arXiv:2507.19881v3 Announce Type: replace-cross 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
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
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2507.19881 [cs.CV]
(or arXiv:2507.19881v3 [cs.CV] for this version)
Text integral sub licența deschisă a publicației · 158 cuvinte
Subiecte conexe
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · 25 sept. 2026
[AINews] not much happened today
Latent Space · 3 oct. 2026
GenSyn10: A Multi-Generative AI Dataset For Benchmarking Image Classification
arXiv · 21 iul. 2026
[AINews] OpenAI reports Navier-Stokes singularity find in 88 hours using Astra-next, roughly 10,000 agents and 130B tokens (>$40M), a contender for second ever Millennium Prize awarded
Latent Space · 9 sept. 2026