The official repository of the paper with supplementary: |

This project is carried out at the Human-Centered AI Lab and the Department of Data Science and AI in the Faculty of Information Technology, Monash University, Melbourne (Clayton), Australia.

Project Members -

Hrishav Bakul Barua (Monash University and TCS Research, Kolkata, India),
Kalin Stefanov (Monash University, Melbourne, Australia),
Lemuel Lai En Che (Monash University, Malaysia),
Abhinav Dhall (Monash University, Melbourne, Australia),
KokSheik Wong (Monash University, Malaysia), and
Ganesh Krishnasami (Monash University, Malaysia).

For any quries kindly contact: hbarua@acm.org/ hrishav.barua@ieee.org

This work is supported by the prestigious Global Excellence and Mobility Scholarship (GEMS), Monash University and a GRS Supplementary Grant Award [Grant No. RMO/L-GRS(SF)/2025-004] from School Postgraduate and Research Committee (SPRC), Monash University. This research is also supported, in part, by the E-Science fund under the project: Innovative High Dynamic Range Imaging - From Information Hiding to Its Applications (Grant No. 01-02-10-SF0327). Funding in the form of Monash IT Student Research (iSR) Scheme 2023 also contributed to this work.

The below image is a qualitative comparison of the proposed CycleHDR method and state-of-the-art method SelfHDR (ICLR'24). We see, our method handles the overexposed portions in the sky more realistically.

Reconstruction of High Dynamic Range (HDR) from Low Dynamic Range (LDR) images is an important computer vision task. There is a significant amount of research utiliz- ing both conventional non-learning methods and modern data- driven approaches, focusing on using both single-exposed and multi-exposed LDR for HDR image reconstruction. However, most current state-of-the-art methods require high-quality paired {LDR;HDR} datasets with limited literature use of unpaired datasets, that is, methods that learn the LDR ↔ HDR mapping between domains. This paper proposes CycleHDR, a method that integrates self-supervision into a modified semantic- and cycle- consistent adversarial architecture that utilizes unpaired LDR and HDR datasets for training. Our method introduces novel artifact- and exposure-aware generators to address visual artifact removal. It also puts forward an encoder and loss to address semantic consistency, another underexplored topic. CycleHDR is the first to use semantic and contextual awareness for the LDR ↔ HDR reconstruction task in a self-supervised setup. The method achieves state-of-the-art performance across several benchmark datasets and reconstructs high-quality HDR images.

Left: Overview of the proposed method architecture where x and y represent LDR and HDR images, respectively. The method is trained with six objectives: adversarial, cycle consistency, identity, heuristic-based, contrastive, and semantic segmentation. G_X and G_Y are the generators while D_X and D_Y are the discriminators. E(.) is the Contrastive Language-Image Pretraining - CLIP encoder. Right: Overview of the proposed generators based on our novel feedback based U-Net architecture. Left part ($${\color{blue}Blue}$$) is the encoder and right part ($${\color{red}Red}$$) is the decoder. The decoder is inside the feedback iteration loop.

Overview of the proposed heuristic-based module. The module outputs the saliency maps for the detected artifacts (in $${\color{red}Red}$$), overexposed pixels (in ($${\color{blue}Blue}$$), and underexposed pixels (in $${\color{yellow}Yellow}$$).

Please check out the paper for more details!!

Depiction of the heuristic-based loss L_heu. The system outputs the the pixel for the detected artifacts, overexposed pixels, and underexposed pixels.

Depiction of the cycle consistency loss L_cyc using an image from the DrTMO dataset.

Left: Depiction of the contrastive loss L_con. Positive ($${\color{green}green}$$) and negative ($${\color{red}Red}$$) pairs in a batch. We use a histogram-equalized version of the LDR processed using the OpenCV function equalizeHist. Right: Depiction of the semantic segmentation loss L_sem. We use Segment Anything (SAM) to generate segmentation classes in the histogram-equalized LDR and reconstructed tone-mapped HDR images.

Please check out the paper for more details!!

I/O: LDR used as input, single-exposed (SE) and multi-exposed (ME), O/P: Reconstructs directly HDR (D) or multi-exposed LDR stack (I), UP: Can be trained with unpaired data, HF: Uses heuristic-based guidance of artifact and exposure infor- mation, Con (Context): Uses local/global image information and relationship among entities in the image, Sem (Semantics): Uses color/texture information and identity of the items in the image, Art (Artifacts): Handles visual artifacts in heavily over/underexposed areas, TM: Supports tone-mapping i.e. HDR -> LDR.

HDR reconstruction (inverse tone-mapping) learned with our self-supervised learning approach. Quantitative comparison with supervised (gray) and unsupervised/weakly-supervised/self- supervised (black) learning methods trained on the paired datasets HDRTV, NTIRE, and HDR-Synth & HDR-Real. LP: Supervised (S), unsupervised (US), weakly- supervised (WS), and self-supervised (SS).

LDR reconstruction (tone-mapping) learned with our self- supervised learning approach. Quantitative comparison with the state-of-the-art one-mapping operators.

Comparison of the SingleHDR(W) U-Net with and without our feedback mechanism on images from the DrTMO dataset. It illustrates the improvement in SingleHDR(W) when we use the proposed feedback U-Net (mod) instead of the original U-Net of SingleHDR(W). The original U-Net produces many artifacts in the output HDR images whereas our modified version with feedback reconstructs artifact-free HDR images.

Please check out the paper for more details!!

You may check some sample visual results.

Note The output/input HDR images (in .hdr format) generated by our pipeline can be viewed using OpenHDRViewer.

Please check out the paper for more details!!

ACM TOG 2017 | HDRCNN - HDR image reconstruction from a single exposure using deep CNNs | Code

ACM TOG 2017 | DrTMO - Deep Reverse Tone Mapping | Code

Eurographics 2018 | ExpandNet - A Deep Convolutional Neural Network for High Dynamic Range Expansion from Low Dynamic Range Content | Code

GlobalSIP 2019 | FHDR - HDR Image Reconstruction from a Single LDR Image using Feedback Network | Code

CVPR 2020 | SingleHDR - Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline | Code

CVPRW 2021 | Two-stage HDR - A two-stage deep network for high dynamic range image reconstruction | Code

IEEE TIP 2021 | HDR-GAN - HDR Image Reconstruction from Multi-Exposed LDR Images with Large Motions | Code

IJCAI 2022 | KUNet - Imaging Knowledge-Inspired Single HDR Image Reconstruction | Code

ECCV 2022 | Ghost-free HDR - Ghost-free High Dynamic Range Imaging with Context-aware Transformer | Code

APSIPA 2023 | ArtHDR-Net - Perceptually Realistic and Accurate HDR Content Creation | Code

ICIP 2024 | HistoHDR-Net - Histogram Equalization for Single LDR to HDR Image Translation| Code

ICCV 2023 | RawHDR - High Dynamic Range Image Reconstruction from a Single Raw Image | Code

IEEE TCSVT 2022 | UPHDR-GAN - Generative Adversarial Network for High Dynamic Range Imaging with Unpaired Data | Code

WACV 2023 | PSENet - Progressive Self-Enhancement Network for Unsupervised Extreme-Light Image Enhancement | Code

WACV 2023 | SingleHDR(W) - Single-Image HDR Reconstruction by Multi-Exposure Generation | Code

ICCV 2023 | GlowGAN-ITM - : Unsupervised Learning of HDR Images from LDR Images in the Wild | Code

ICLR 2024 | SelfHDR - Self-Supervised High Dynamic Range Imaging with Multi-Exposure Images in Dynamic Scenes | Code

VPQM 2015 | HDR-Eye - Visual attention in LDR and HDR images | Dataset

ACM TOG 2017 | Kalantari et al. - Deep high dynamic range imaging of dynamic scenes | Dataset

IEEE Access 2020 | LDR-HDR Pair - Dynamic range expansion using cumulative histogram learning for high dynamic range image generation | Dataset

CVPR 2020 | HDR-Synth & HDR-Real - Single-image HDR reconstruction by learning to reverse the camera pipeline | Dataset

ICCV 2021 | HDRTV - A New Journey from SDRTV to HDRTV | Dataset

CVPR 2021 | NTIRE - Ntire 2021 challenge on high dynamic range imaging: Dataset, methods and results.| Dataset

ACM TOG 2017 | DrTMO - Deep Reverse Tone Mapping | Dataset

WACV 2025 | GTA-HDR - A Large-Scale Synthetic Dataset for HDR Image Reconstruction | Dataset

μ-law operator | Link

Reinhard's operator | Link

Photomatix | Link

GlowGAN-prior | Link

PSNR - Peak Signal to Noise Ratio | Link

SSIM - Structural Similarity Index Measure | Link

LPIPS - Learned Perceptual Image Patch Similarity | Link

HDR-VDP-2 - High Dynamic Range Visual Differences Predictor | Link

If you find our work (i.e., the code, the theory/concept, or the dataset) useful for your research or development activities, please consider citing our work as follows:

@article{barua2024cycle,
  title={A Cycle Ride to HDR: Semantics Aware Self-Supervised Framework for Unpaired LDR-to-HDR Image Translation},
  author={Barua, Hrishav Bakul and Kalin, Stefanov and Che, Lemuel Lai En and Abhinav, Dhall and KokSheik, Wong and Ganesh, Krishnasamy},
  journal={arXiv preprint arXiv:2410.15068},
  year={2024}
}

@article{barua2025physhdr,
  title={PhysHDR: When Lighting Meets Materials and Scene Geometry in HDR Reconstruction},
  author={Barua, Hrishav Bakul and Stefanov, Kalin and Krishnasamy, Ganesh and Wong, KokSheik and Dhall, Abhinav},
  journal={arXiv preprint arXiv:2509.16869},
  year={2025}
}

@InProceedings{Barua_2025_WACV,
    author    = {Barua, Hrishav Bakul and Stefanov, Kalin and Wong, KokSheik and Dhall, Abhinav and Krishnasamy, Ganesh},
    title     = {GTA-HDR: A Large-Scale Synthetic Dataset for HDR Image Reconstruction},
    booktitle = {Proceedings of the Winter Conference on Applications of Computer Vision (WACV)},
    month     = {February},
    year      = {2025},
    pages     = {7865-7875}
}

Related works:

@inproceedings{barua2023arthdr,
  title={ArtHDR-Net: Perceptually Realistic and Accurate HDR Content Creation},
  author={Barua, Hrishav Bakul and Krishnasamy, Ganesh and Wong, KokSheik and Stefanov, Kalin and Dhall, Abhinav},
  booktitle={2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)},
  pages={806--812},
  year={2023},
  organization={IEEE}
}

@inproceedings{barua2024histohdr,
  title={HistoHDR-Net: Histogram equalization for single LDR to HDR image translation},
  author={Barua, Hrishav Bakul and Krishnasamy, Ganesh and Wong, KokSheik and Dhall, Abhinav and Stefanov, Kalin},
  booktitle={2024 IEEE International Conference on Image Processing (ICIP)},
  pages={2730--2736},
  year={2024},
  organization={IEEE}
}

----------------------------------------------------------------------------------------
Copyright 2025 | All the authors and contributors of this repository as mentioned above.
----------------------------------------------------------------------------------------

Please check the License Agreement.