Research map / Architectures and efficiency / CNN design
MobileIE: An Extremely Lightweight and Effective ConvNet for Real-Time Image Enhancement on Mobile Devices
Hailong Yan, Ao Li, Xiangtao Zhang, Zhe Liu, Zenglin Shi, Ce Zhu, Le Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use LVW loss instead of L1 loss (ablation) | MobileIE · UIEB | PSNR: 22.20 -> 22.57 | Improved |
| use LVW loss instead of Charbonnier loss (ablation) | MobileIE · UIEB | PSNR: 22.31 -> 22.57 | Improved |
| use MBRConv instead of RepVGG (ablation) | MobileIE · LOLv1+LOLv2-Real | PSNR: 22.69 -> 24.35 | Improved |
| use MBRConv instead of ECBSR (ablation) | MobileIE · LOLv1+LOLv2-Real | PSNR: 23.96 -> 24.35 | Improved |
| add incremental weight optimization (IWO) (ablation) | MobileIE MBRConv · LOLv1+LOLv2-Real | PSNR: 24.02 -> 24.35 | Improved |
| add feature self-transform (FST) (ablation) | MobileIE · LOLv1+LOLv2-Real | PSNR: 21.12 -> 22.60 | Improved |
| use FST instead of MUL feature transform (ablation) | MobileIE · LOLv1+LOLv2-Real | PSNR: 21.80 -> 22.60 | Improved |
| use HDPA instead of SE-Net attention (ablation) | MobileIE · LOLv1 | PSNR: 22.27 -> 23.26 | Improved |
| use HDPA instead of CBAM attention (ablation) | MobileIE · LOLv1 | PSNR: 22.38 -> 23.26 | Improved |
| add multi-branch training (vs inference network only) (ablation) | MobileIE · UIEB | PSNR: 21.48 -> 22.57 | Improved |
| use IWO and LVW at inference (ablation) | MobileIE · efficiency | latency: 0.895 ms unchanged | No clear effect |
Closest papers by meaning
- RepViT: Revisiting Mobile CNN From ViT Perspective (CVPR 2024 · similarity 0.87)
- IFORMER: INTEGRATING CONVNET AND TRANSFORMER FOR MOBILE APPLICATION (ICLR 2025 · similarity 0.85)
- MobileMamba: Lightweight Multi-Receptive Visual Mamba Network (CVPR 2025 · similarity 0.85)
- Real-time Core-Periphery Guided ViT with Smart Data Layout Selection on Mobile Devices (NeurIPS 2024 · similarity 0.84)
- SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training (CVPR 2025 · similarity 0.84)
- Real-World Mobile Image Denoising Dataset with Efficient Baselines (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
