{"id":25118040,"url":"https://github.com/retkowsky/colorizer","last_synced_at":"2025-04-02T12:30:03.601Z","repository":{"id":61962529,"uuid":"555302601","full_name":"retkowsky/colorizer","owner":"retkowsky","description":"DeOldify is a projet to colorize and restore old images and film footage","archived":false,"fork":false,"pushed_at":"2022-10-24T09:48:30.000Z","size":47713,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-02-08T03:38:55.198Z","etag":null,"topics":["ai","azure","azure-ml","computer-vision","deoldify"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","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/retkowsky.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}},"created_at":"2022-10-21T10:10:46.000Z","updated_at":"2024-12-13T00:51:08.000Z","dependencies_parsed_at":"2023-01-20T12:04:40.776Z","dependency_job_id":null,"html_url":"https://github.com/retkowsky/colorizer","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/retkowsky%2Fcolorizer","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/retkowsky%2Fcolorizer/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/retkowsky%2Fcolorizer/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/retkowsky%2Fcolorizer/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/retkowsky","download_url":"https://codeload.github.com/retkowsky/colorizer/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":246815080,"owners_count":20838374,"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":["ai","azure","azure-ml","computer-vision","deoldify"],"created_at":"2025-02-08T03:39:07.356Z","updated_at":"2025-04-02T12:30:03.571Z","avatar_url":"https://github.com/retkowsky.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# DeOldify is a projet to colorize and restore old images and film footage.\n\u003e All credits: https://github.com/jantic/DeOldify\n\nDeOldify is a **Black and White image Colorizing library created by Jason Antic**. DeOldify introduced the NoGAN technique to solve some of the key problems to make hyper-realistic colorization images and video. DeOldify provides **three primary models for different use case**.\n\n## Some colorized images\n\u003cimg src=\"Présentation1.gif\"\u003e\n\n## A colorized video\n\u003ca href=\"https://github.com/retkowsky/colorizer/raw/main/colorized_video_train_en_gare_de_la_ciotat.mp4\"\u003eDemo Video\u003ca\u003e\n\n## Python notebook\n\u003ca href=\"Image Colorizer.ipynb\"\u003eImage Colorizer.ipynb\u003ca\u003e\n\n## Models\nEach of these has some limitations and benefit:\n\n#### Artistic Model\nThis model achieves **vibrant color and detailed images**, but you have to adjust the parameters a lot to get the best results. You have to adjust the rendering resolution and factors to get the most accurate colorize image.\n\nThe model uses a **resnet34 backbone on a UNet** with an emphasis on the depth of layers on the decoder side. And it is trained on five critic pretrain/GAN cycle repeats via NoGAN.\n\n#### Stable Model \nThis model archives the **best results in portraits and landscapes**. It makes sure nothing got too much colored and makes the most part of the image remain gray like faces and limbs. It is less hyper-realistic but it makes sure nothing looks more colored.\n\nIt uses a **resnet101 backbone on a UNet** with an emphasis on the width of layers on the decoder side.\n\n#### Video Model\nAs the name suggests, it is a model used to **color the videos** and we are going to see each of these models working in a python environment. It gives smooth, consistent, and flicker-free Video. This model is the same as a stable model in the case of architecture but different in training. **DeOldify is trained on 2.2% of Imagenet dataset once at 192px, using only the initial generator/critic pretrain/GAN NoGAN training.**\n\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fretkowsky%2Fcolorizer","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fretkowsky%2Fcolorizer","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fretkowsky%2Fcolorizer/lists"}