ALLIE: Autoencoder-based Low-Light Image Enhancement

Le résumé fourni par la source

Operating in low-light environments remains one of the most critical challenges for autonomous robotic systems. Under such conditions, image quality is often compromised by noise, reduced detail, and low visibility, severely limiting the system's ability to perceive and interact with the environment. In this work, we propose ALLIE, an innovative deep neural network model based on an Autoencoder architecture, designed specifically for enhancing images taken in low-light conditions. Unlike traditional methods that rely on manual adjustment of brightness, contrast, or noise filtering, ALLIE performs enhancement automatically through end-to-end learning. The network learns to transform dark images into well-exposed versions by capturing complex image representations that improve clarity while preserving important scene details. We evaluated the proposed method on the paired Low-Light (LOL) image dataset, in both versions v1 and v2, using reference-based quality metrics such as PSNR and SSIM to quantify fidelity between enhanced images and their corresponding groundtruths. Experimental results demonstrate that ALLIE outperforms conventional and state-of-the-art approaches, producing images with superior visual quality, which are suitable for diverse applications in robotics and computer vision.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.