analiza aiqbench
Multinex addresses the practical constraints of edge deployment in computer vision by introducing an ultra-lightweight framework that outperforms larger state-of-the-art models while maintaining low computational costs. By leveraging a multi-prior Retinex formulation, it effectively resolves exposure and color artifacts common in existing low-light image enhancement techniques without relying on massive parameter counts. This research highlights the growing trend of optimizing neural architectures for efficiency rather than just raw scale, making advanced vision tasks viable on resource-constrained devices.
Multinex: Lightweight Low-light Image Enhancement via Multi-prior Retinex
arXiv:2604.10359v3 Announce Type: replace-cross Abstract: Low-light image enhancement (LLIE) aims to restore natural visibility, color fidelity, and structural detail under severe illumination degradation. State-of-the-art (SOTA) LLIE techniques often rely on large models and multi-stage training, limiting practicality for edge deployment. Moreover, their dependence on a single color space introduces instability and visible exposure or color artifacts. To address these, we propose Multinex, an ultra-lightweight structured framework that integrates multiple fine-grained representations within a principled Retinex residual formulation. It decomposes an image into illumination and color prior stacks derived from distinct analytic representations, and learns to fuse these representations into luminance and reflectance adjustments required to correct exposure. By prioritizing enhancement over reconstruction and exploiting lightweight neural operations, Multinex significantly reduces computational cost, exemplified by its lightweight (45K parameters) and nano (0.7K parameters) versions. Extensive benchmarks show that all lightweight variants significantly outperform their corresponding lightweight SOTA models, and reach comparable performance to heavy models. Paper page available at https://albrateanu.github.io/multinex.
Abstract:Low-light image enhancement (LLIE) aims to restore natural visibility, color fidelity, and structural detail under severe illumination degradation. State-of-the-art (SOTA) LLIE techniques often rely on large models and multi-stage training, limiting practicality for edge deployment. Moreover, their dependence on a single color space introduces instability and visible exposure or color artifacts. To address these, we propose Multinex, an ultra-lightweight structured framework that integrates multiple fine-grained representations within a principled Retinex residual formulation. It decomposes an image into illumination and color prior stacks derived from distinct analytic representations, and learns to fuse these representations into luminance and reflectance adjustments required to correct exposure. By prioritizing enhancement over reconstruction and exploiting lightweight neural operations, Multinex significantly reduces computational cost, exemplified by its lightweight (45K parameters) and nano (0.7K parameters) versions. Extensive benchmarks show that all lightweight variants significantly outperform their corresponding lightweight SOTA models, and reach comparable performance to heavy models. Paper page available at this https URL.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2604.10359 [cs.CV]
(or arXiv:2604.10359v3 [cs.CV] for this version)
Text integral sub licența deschisă a publicației · 173 cuvinte
Subiecte conexe
Researchers unlock quantum infrared sensing with a new screen
Quantum Zeitgeist · 13 sept. 2026
Combining scEEG and PPG for reliable sleep staging
arXiv · 15 oct. 2026
Physen-Noise2Noise: Physics-Guided Self-Supervised Defocus Deblurring with Bias Correction under Low-Light Conditions
arXiv · 4 aug. 2026
The Machines that Make the Machines
NVIDIA · 7 oct. 2026