Illuminating Autonomy Federated Learning for Object Detection in Autonomous Vehicles Under Low Light
Year: 2024
Type: article
Abstract: Autonomous vehicles rely heavily on object detection systems to navigate safely, yet low-light conditions pose significant challenges to these systems. Traditional centralized machine learning approaches require substantial data aggregation, raising privacy and security concerns. This paper introduces an innovative approach, Illuminating Autonomy Federated Learning (IAFL), designed to enhance object detection in autonomous vehicles operating under low-light conditions. IAFL leverages federated learning to train object detection models across multiple vehicles without centralizing the data, thus preserving privacy while utilizing diverse realworld driving data. The approach integrates advanced image enhancement techniques to improve low-light object visibility, combined with robust federated learning algorithms to ensure model accuracy and generalization across varying environments. Extensive simulations and real-world testing demonstrate that IAFL significantly improves detection performance in low-light scenarios, maintaining high accuracy and efficiency while ensuring data privacy and security. This research marks a pivotal step toward safer and more reliable autonomous driving in challenging lighting conditions.
Source: Journal of engineering sciences.
Institution: Malla Reddy Engineering College for Women
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4403432550