{"id":20881944,"url":"https://github.com/aiff22/microisp","last_synced_at":"2025-04-11T01:02:44.816Z","repository":{"id":244461243,"uuid":"815270540","full_name":"aiff22/MicroISP","owner":"aiff22","description":null,"archived":false,"fork":false,"pushed_at":"2024-06-14T20:45:42.000Z","size":49308,"stargazers_count":8,"open_issues_count":2,"forks_count":2,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-11T01:02:31.691Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/aiff22.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-06-14T18:19:15.000Z","updated_at":"2025-03-06T08:23:10.000Z","dependencies_parsed_at":"2024-06-14T21:55:46.964Z","dependency_job_id":null,"html_url":"https://github.com/aiff22/MicroISP","commit_stats":null,"previous_names":["aiff22/microisp"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aiff22%2FMicroISP","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aiff22%2FMicroISP/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aiff22%2FMicroISP/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aiff22%2FMicroISP/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/aiff22","download_url":"https://codeload.github.com/aiff22/MicroISP/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248322597,"owners_count":21084336,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-11-18T07:27:08.846Z","updated_at":"2025-04-11T01:02:44.785Z","avatar_url":"https://github.com/aiff22.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"## MicroISP: Processing 32MP Photos on Mobile Devices with Deep Learning\n\n\u003cbr/\u003e\n\n\u003cimg src=\"https://people.ee.ethz.ch/~ihnatova/demo_microisp/MicroISP_teaser.jpg\"/\u003e\n\n#### 1. Overview [[Paper]](https://arxiv.org/pdf/2211.06770) [[Project Webpage]](http://people.ee.ethz.ch/~ihnatova/microisp.html)\n\nThis repository provides the implementation of the RAW-to-RGB mapping approach and MicroISP CNN presented in [this paper](https://arxiv.org/pdf/2211.06770). The model is trained to convert **RAW Bayer data** obtained directly from mobile camera sensor into photos captured with a professional medium format 102MP [Fujifilm GFX100](https://www.dpreview.com/reviews/fujifilm-gfx-100-review) camera, thus replacing the entire hand-crafted ISP camera pipeline. The provided pre-trained MicroISP model can be used to generate full-resolution **32MP photos** from RAW (DNG) image files captured using the Sony Exmor IMX586 camera sensor directly on mobile devices.\n\n\u003cbr/\u003e\n\n#### 2. Prerequisites\n\n- Python: scipy, numpy, imageio and pillow packages\n- [TensorFlow 2.X](https://www.tensorflow.org/install/)\n- [Optioinal] Nvidia GPU  + [CUDA cuDNN](https://developer.nvidia.com/cudnn)\n\n\u003cbr/\u003e\n\n#### 3. MicroISP CNN\n\n\u003cbr/\u003e\n\n\u003cimg src=\"https://people.ee.ethz.ch/~ihnatova/demo_microisp/microisp_architecture.png\" alt=\"drawing\" width=\"1000\"/\u003e\n\n\u003cbr/\u003e\n\nThe model accepts the raw RGBG Bayer data coming directly from the camera sensor. The input is then grouped in 4 feature maps corresponding to each of the four RGBG color channels using the space-to-depth op. Next, this input is processed in parallel in 3 model branches corresponding to the R, G and B color channels and consisting of N residual building blocks. After applying the depth-to-space op at the end of each branch, their outputs are concatenated into the reconstructed RGB photo.\n\nThe proposed MicroISP model contains only layers supported by the **Neural Networks API 1.2**, and thus can run on any NNAPI-compliant AI accelerator (such as NPU, APU, DSP or GPU) available on mobile devices with Android 10 and above. The size of the MicroISP network is only **158 KB** when exported for inference using the TFLite FP32 format. The model consumes around **90**, **475** and **975MB** of RAM when processing **FullHD**, **12MP** and **32MP** photos on mobile GPUs, respectively. Its GPU runtimes on various platforms for images of different resolutions are provided below: \n\n\u003cbr/\u003e\n\n\u003cimg src=\"https://people.ee.ethz.ch/~ihnatova/demo_microisp/MicroISP_Runtime.png\" alt=\"drawing\" width=\"1000\"/\u003e\n\n\u003cbr/\u003e\n\n#### 4. Test the provided pre-trained models on full-resolution RAW image files\n\n```bash\npython inference.py\n```\n\nThe model will then process DNG images from the ``sample_RAW_photos`` directory and save the resulting RGB/PNG images to the ``sample_visual_results`` folder.\n\n\u003cbr/\u003e\n\n#### 5. Folder structure\n\n\u003e```pretrained_weights/```   \u0026nbsp; - \u0026nbsp; the folder with the provided pre-trained MicroISP model \u003cbr/\u003e\n\u003e```sample_RAW_photos/```        \u0026nbsp; - \u0026nbsp; the folder with sample RAW/DNG images from the Fujifilm UltraISP Dataset \u003cbr/\u003e\n\u003e```model.py```           \u0026nbsp; - \u0026nbsp; TensorFlow MicroISP implementation \u003cbr/\u003e\n\u003e```inference.py```     \u0026nbsp; - \u0026nbsp; applying the pre-trained model to full-resolution test images \u003cbr/\u003e\n\u003e```export_to_tflite.py```      \u0026nbsp; - \u0026nbsp; model export to TensorFlow Lite format for on-device deployment \u003cbr/\u003e\n\n\u003cbr/\u003e\n\n#### 6. License\n\nCopyright (C) 2024 Andrey Ignatov. All rights reserved.\n\nLicensed under the [CC BY-NC-SA 4.0 (Attribution-NonCommercial-ShareAlike 4.0 International)](https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode).\n\nThe code is released for academic research use only.\n\n\u003cbr/\u003e\n\n#### 7. Citation\n\n```\n@inproceedings{ignatov2022microisp,\n  title={MicroISP: Processing 32MP Photos on Mobile Devices with Deep Learning},\n  author={Ignatov, Andrey and Sycheva, Anastasia and Timofte, Radu and Tseng, Yu and Xu, Yu-Syuan and Yu, Po-Hsiang and Chiang, Cheng-Ming and Kuo, Hsien-Kai and Chen, Min-Hung and Cheng, Chia-Ming and others},\n  booktitle={European Conference on Computer Vision},\n  pages={729--746},\n  year={2022},\n  organization={Springer}\n}\n```\n\u003cbr/\u003e\n\n#### 8. Any further questions?\n\n```\nPlease contact Andrey Ignatov (andrey@vision.ee.ethz.ch) for more information\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faiff22%2Fmicroisp","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faiff22%2Fmicroisp","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faiff22%2Fmicroisp/lists"}