dorsal/arxiv
View SchemaBridging Robustness and Efficiency: Real-Time Low-Light Enhancement via Attention U-Net GAN
| Authors | Yash Thesia, Meera Suthar |
|---|---|
| Categories | |
| ArXiv ID | 2601.06518vv1 |
| URL | https://arxiv.org/abs/2601.06518 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
Recent advancements in Low-Light Image Enhancement (LLIE) have focused heavily on Diffusion Probabilistic Models, which achieve high perceptual quality but suffer from significant computational latency (often exceeding 2-4 seconds per image). Conversely, traditional CNN-based baselines offer real-time inference but struggle with "over-smoothing," failing to recover fine structural details in extreme low-light conditions. This creates a practical gap in the literature: the lack of a model that provides generative-level texture recovery at edge-deployable speeds. In this paper, we address this trade-off by proposing a hybrid Attention U-Net GAN. We demonstrate that the heavy iterative sampling of diffusion models is not strictly necessary for texture recovery. Instead, by integrating Attention Gates into a lightweight U-Net backbone and training within a conditional adversarial framework, we can approximate the high-frequency fidelity of generative models in a single forward pass. Extensive experiments on the SID dataset show that our method achieves a best-in-class LPIPS score of 0.112 among efficient models, significantly outperforming efficient baselines (SID, EnlightenGAN) while maintaining an inference latency of 0.06s. This represents a 40x speedup over latent diffusion models, making our approach suitable for near real-time applications.
{
"annotation_id": "da281791-7de8-4b37-b023-268e1da9c4dc",
"date_created": "2026-02-17T05:53:08.711000Z",
"date_modified": "2026-02-17T05:53:08.711000Z",
"file_hash": "60a696fb4733da57f35cb9c9837748850752b626a067a25d769fcc048bd49297",
"private": false,
"record": {
"abstract": "Recent advancements in Low-Light Image Enhancement (LLIE) have focused heavily on Diffusion Probabilistic Models, which achieve high perceptual quality but suffer from significant computational latency (often exceeding 2-4 seconds per image). Conversely, traditional CNN-based baselines offer real-time inference but struggle with \"over-smoothing,\" failing to recover fine structural details in extreme low-light conditions. This creates a practical gap in the literature: the lack of a model that provides generative-level texture recovery at edge-deployable speeds. In this paper, we address this trade-off by proposing a hybrid Attention U-Net GAN. We demonstrate that the heavy iterative sampling of diffusion models is not strictly necessary for texture recovery. Instead, by integrating Attention Gates into a lightweight U-Net backbone and training within a conditional adversarial framework, we can approximate the high-frequency fidelity of generative models in a single forward pass. Extensive experiments on the SID dataset show that our method achieves a best-in-class LPIPS score of 0.112 among efficient models, significantly outperforming efficient baselines (SID, EnlightenGAN) while maintaining an inference latency of 0.06s. This represents a 40x speedup over latent diffusion models, making our approach suitable for near real-time applications.",
"arxiv_id": "2601.06518",
"authors": [
"Yash Thesia",
"Meera Suthar"
],
"categories": [
"cs.CV"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "Bridging Robustness and Efficiency: Real-Time Low-Light Enhancement via Attention U-Net GAN",
"url": "https://arxiv.org/abs/2601.06518",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "d5395cb0-cb63-480e-b1fa-066d345628fd",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}