# SAFNet: Selective Alignment Fusion Network for Efficient HDR Imaging

arXiv: https://arxiv.org/abs/2407.16308 · HTML: https://mlsanity.com/paper/2407.16308/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- Builds on 0 of its 48 references that are in the map.
- First posted on arXiv 23 Jul 2024.

## Built on by (most cited first, 23 of 23)

- [LEDiff: Latent Exposure Diffusion for HDR Generation](https://mlsanity.com/paper/2412.14456/) (2024)
- [UltraFusion: Ultra High Dynamic Imaging using Exposure Fusion](https://mlsanity.com/paper/2501.11515/) (2025)
- [Perspective-Invariant 3D Object Detection](https://mlsanity.com/paper/2507.17665/) (2025)
- [GaussHDR: High Dynamic Range Gaussian Splatting via Learning Unified 3D and 2D Local Tone Mapping](https://mlsanity.com/paper/2503.10143/) (2025)
- [X2HDR: HDR Image Generation in a Perceptually Uniform Space](https://mlsanity.com/paper/2602.04814/) (2026)
- [It Takes Two: A Duet of Periodicity and Directionality for Burst Flicker Removal](https://mlsanity.com/paper/2603.22794/) (2026)
- [Capturing Stable HDR Videos Using a Dual-Camera System](https://mlsanity.com/paper/2507.06593/) (2025)
- [AdaptiveAE: An Adaptive Exposure Strategy for HDR Capturing in Dynamic Scenes](https://mlsanity.com/paper/2508.13503/) (2025)
- [DiffHDR: Re-Exposing LDR Videos with Video Diffusion Models](https://mlsanity.com/paper/2604.06161/) (2026)
- [Boosting HDR Image Reconstruction via Semantic Knowledge Transfer](https://mlsanity.com/paper/2503.15361/) (2025)
- [S2R-HDR: A Large-Scale Rendered Dataset for HDR Fusion](https://mlsanity.com/paper/2504.07667/) (2025)
- [F2HDR: Two-Stage HDR Video Reconstruction via Flow Adapter and Physical Motion Modeling](https://mlsanity.com/paper/2603.14920/) (2026)
- [GMODiff: One-Step Gain Map Refinement with Diffusion Priors for HDR Reconstruction](https://mlsanity.com/paper/2512.16357/) (2025)
- [Physically Inspired Gaussian Splatting for HDR Novel View Synthesis](https://mlsanity.com/paper/2603.28020/) (2026)
- [LoCAtion: Long-time Collaborative Attention Framework for High Dynamic Range Video Reconstruction](https://mlsanity.com/paper/2603.14377/) (2026)
- [Lucky High Dynamic Range Smartphone Imaging](https://mlsanity.com/paper/2604.19976/) (2026)
- [Single-Shot HDR Recovery via a Video Diffusion Prior](https://mlsanity.com/paper/2605.11628/) (2026)
- [Generating HDR Video from SDR Video](https://mlsanity.com/paper/2605.14703/) (2026)
- [ExpoMotion: A Large-Scale Benchmark and A Householder Projection Network for Multi-Exposure Fusion](https://mlsanity.com/paper/2607.03110/) (2026)
- [Coarse-to-fine Framework for Generative MEF via Implicit Neural Representation](https://mlsanity.com/paper/2607.17611/) (2026)
- [Dual-Output Multi-Exposure HDR Reconstruction via SDR Fusion and Gain Map Inverse Tone Mapping](https://mlsanity.com/paper/2608.05626/) (2026)
- [Depth-guided Multi-view Exposure Bracketing for HDR Robot Vision](https://mlsanity.com/paper/2608.16014/) (2026)
- [High Dynamic Range Video Reconstruction from Single-Exposure Raw Sequences](https://mlsanity.com/paper/2609.27274/) (2026)

## Abstract (arXiv)

Multi-exposure High Dynamic Range (HDR) imaging is a challenging task when facing truncated texture and complex motion. Existing deep learning-based methods have achieved great success by either following the alignment and fusion pipeline or utilizing attention mechanism. However, the large computation cost and inference delay hinder them from deploying on resource limited devices. In this paper, to achieve better efficiency, a novel Selective Alignment Fusion Network (SAFNet) for HDR imaging is proposed. After extracting pyramid features, it jointly refines valuable area masks and cross-exposure motion in selected regions with shared decoders, and then fuses high quality HDR image in an explicit way. This approach can focus the model on finding valuable regions while estimating their easily detectable and meaningful motion. For further detail enhancement, a lightweight refine module is introduced which enjoys privileges from previous optical flow, selection masks and initial prediction. Moreover, to facilitate learning on samples with large motion, a new window partition cropping method is presented during training. Experiments on public and newly developed challenging datasets show that proposed SAFNet not only exceeds previous SOTA competitors quantitatively and qualitatively, but also runs order of magnitude faster. Code and dataset is available at https://github.com/ltkong218/SAFNet.