Illuminating Darkness: Learning to Enhance Low-light Images In-the-Wild
This work addresses the problem of enhancing low-light images in real-world conditions for computer vision applications, representing a strong specific gain rather than a foundational advancement.
The paper tackles the challenge of single-shot low-light image enhancement by introducing a large-scale dataset (LSD) with 6,425 paired images and proposing TFFormer, a hybrid model that achieves state-of-the-art results, including a +2.45 dB PSNR improvement on LSD and +6.80 mAP on low-light object detection.
Single-shot low-light image enhancement (SLLIE) remains challenging due to the limited availability of diverse, real-world paired datasets. To bridge this gap, we introduce the Low-Light Smartphone Dataset (LSD), a large-scale, high-resolution (4K+) dataset collected in the wild across a wide range of challenging lighting conditions (0.1 to 200 lux). LSD contains 6,425 precisely aligned low and normal-light image pairs, selected from over 8,000 dynamic indoor and outdoor scenes through multi-frame acquisition and expert evaluation. To evaluate generalization and aesthetic quality, we collect 2,117 unpaired low-light images from previously unseen devices. To fully exploit LSD, we propose TFFormer, a hybrid model that encodes luminance and chrominance (LC) separately to reduce color-structure entanglement. We further propose a cross-attention-driven joint decoder for context-aware fusion of LC representations, along with LC refinement and LC-guided supervision to significantly enhance perceptual fidelity and structural consistency. TFFormer achieves state-of-the-art results on LSD (+2.45 dB PSNR) and substantially improves downstream vision tasks, such as low-light object detection (+6.80 mAP on ExDark).