Research map / Image generation / Image restoration
DeepFuse: A Deep Unsupervised Approach for Exposure Fusion With Extreme Exposure Image Pairs
K. Ram Prabhakar, V Sai Srikar, R. Venkatesh Babu
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 |
|---|---|---|---|
| fuse features by addition instead of product (ablation) | DeepFuse · 23 test images | average MEF SSIM: 0.8210 -> 0.9782 | Improved |
| fuse features by addition instead of concatenation (ablation) | DeepFuse · 23 test images | average MEF SSIM: 0.9430 -> 0.9782 | Improved |
| fuse features by addition instead of max (ablation) | DeepFuse · 23 test images | average MEF SSIM: 0.9638 -> 0.9782 | Improved |
| fuse features by addition instead of mean (ablation) | DeepFuse · 23 test images | average MEF SSIM: 0.9750 -> 0.9782 | No clear effect |
| train with MEF SSIM unsupervised loss instead of imitating Mertens/GFF outputs (ablation) | DeepFuse · MEF SSIM test sequences | MEF SSIM: DF-Baseline 0.9477 -> DF-UnSupervised 0.9813 on AgiaGalini | Improved |
| use SSIM loss instead of l2 for supervised baseline (ablation) | DF-Baseline · test sequences | visual quality | Improved |
| tie weights of the two pre-fusion channels | DeepFuse · test sequences | parameters/convergence | Improved |
| use DeepFuse with 3 input images instead of Mertens | DeepFuse · 3-image sequences | average MEF SSIM: 0.979 -> 0.987 | Improved |
| use DeepFuse with 4 input images instead of Mertens | DeepFuse · 4-image sequences | average MEF SSIM: 0.978 -> 0.972 | Hurt |
| run DeepFuse instead of Mertens | DeepFuse · 1920*1200 image pair | running time (s): 2.76 -> 0.82 | Improved |
Later papers that compared with this method
3 later accepted papers report DeepFuse as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Hybrid-Supervised Dual-Search: Leveraging Automatic Learning for Loss-Free Multi-Exposure Image Fusion (AAAI 2024 · better on 1 of 1 shared result)
- Unsupervised Multi-Exposure Image Fusion Breaking Exposure Limits via Contrastive Learning (AAAI 2023 · better on 2 of 3 shared results)
- FusionDN: A Unified Densely Connected Network for Image Fusion (AAAI 2020 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TransMEF: A Transformer-Based Multi-Exposure Image Fusion Framework Using Self-Supervised Multi-Task Learning (AAAI 2022 · similarity 0.88)
- Unsupervised Multi-Exposure Image Fusion Breaking Exposure Limits via Contrastive Learning (AAAI 2023 · similarity 0.87)
- MEFLUT: Unsupervised 1D Lookup Tables for Multi-exposure Image Fusion (ICCV 2023 · similarity 0.87)
- Hybrid-Supervised Dual-Search: Leveraging Automatic Learning for Loss-Free Multi-Exposure Image Fusion (AAAI 2024 · similarity 0.87)
- U2Fusion: A Unified Unsupervised Image Fusion Network (TPAMI 2020 · similarity 0.87)
- UltraFusion: Ultra High Dynamic Imaging using Exposure Fusion (CVPR 2025 · 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.
