Exploring Personalized Federated Learning Architectures for Violence Detection in Surveillance Videos

2025-04-01

AI Summary

Urban surveillance videos pose significant challenges for violent incident detection, including massive data volume, multimodal inputs, non-independent-and-identically-distributed (Non-IID) data across distributed nodes, and severe class imbalance. To address these issues while preserving data privacy, this paper proposes a personalized federated learning (PFL)-based framework. It pioneers the adaptation of the Flower federated system—enhanced with node-specific personalization layers—to video behavior classification, enabling each surveillance node to maintain a customized local model without sharing raw data. This design effectively mitigates data heterogeneity and Non-IID effects while ensuring strict on-device data confinement. Extensive experiments demonstrate state-of-the-art performance: up to 99.3% accuracy on both balanced and imbalanced benchmarks, alongside substantial improvements in model generalizability, training efficiency, and system scalability. The framework establishes a novel paradigm for large-scale, privacy-preserving, distributed intelligent security systems.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Adversarial Attacks & Robustness
• Search and Optimization: Distributed Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Security and Privacy: Data transparency and provenance

Abstract

The challenge of detecting violent incidents in urban surveillance systems is compounded by the voluminous and diverse nature of video data. This paper presents a targeted approach using Personalized Federated Learning (PFL) to address these issues, specifically employing the Federated Learning with Personalization Layers method within the Flower framework. Our methodology adapts learning models to the unique data characteristics of each surveillance node, effectively managing the heterogeneous and non-IID nature of surveillance video data. Through rigorous experiments conducted on balanced and imbalanced datasets, our PFL models demonstrated enhanced accuracy and efficiency, achieving up to 99.3% accuracy. This study underscores the potential of PFL to significantly improve the scalability and effectiveness of surveillance systems, offering a robust, privacy-preserving solution for violence detection in complex urban environments.

Problem

• Detecting violence in diverse urban surveillance videos
• Addressing data heterogeneity with personalized federated learning
• Improving accuracy and privacy in violence detection systems

Innovation

• Personalized Federated Learning for violence detection
• Federated Learning with Personalization Layers
• Adapts models to heterogeneous surveillance data