{"id":26778564,"url":"https://github.com/mat-kubiak/hybrid-cgiqa","last_synced_at":"2026-05-16T22:05:20.224Z","repository":{"id":271931176,"uuid":"896943994","full_name":"mat-kubiak/Hybrid-CGIQA","owner":"mat-kubiak","description":"A Lightweight NR-IQA Model for CGIs Based on the CGIQA-6K Dataset","archived":false,"fork":false,"pushed_at":"2025-04-14T10:33:51.000Z","size":11977,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-14T11:38:57.713Z","etag":null,"topics":["cgi","computer-generated-imagery","image-quality-assessment","iqa","tensorflow","tensorflow-models"],"latest_commit_sha":null,"homepage":"","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/mat-kubiak.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"licenses/LICENSE","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,"zenodo":null}},"created_at":"2024-12-01T17:28:08.000Z","updated_at":"2025-04-14T10:33:55.000Z","dependencies_parsed_at":"2025-01-27T07:23:24.205Z","dependency_job_id":"d259192e-252b-4dab-af7f-61b8abc325f0","html_url":"https://github.com/mat-kubiak/Hybrid-CGIQA","commit_stats":null,"previous_names":["mat-kubiak/iqa-thesis","mat-kubiak/hybrid-cgiqa"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/mat-kubiak/Hybrid-CGIQA","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FHybrid-CGIQA","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FHybrid-CGIQA/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FHybrid-CGIQA/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FHybrid-CGIQA/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mat-kubiak","download_url":"https://codeload.github.com/mat-kubiak/Hybrid-CGIQA/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FHybrid-CGIQA/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":268385731,"owners_count":24242101,"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","status":"online","status_checked_at":"2025-08-02T02:00:12.353Z","response_time":74,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"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":["cgi","computer-generated-imagery","image-quality-assessment","iqa","tensorflow","tensorflow-models"],"created_at":"2025-03-29T05:37:36.535Z","updated_at":"2026-05-16T22:05:20.183Z","avatar_url":"https://github.com/mat-kubiak.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Hybrid-CGIQA\n\nA lightweight NR-IQA model for CGIs based on the CGIQA-6K dataset and a hybrid architecture.\n\nThis code and the published model is a part the engineering thesis \"Subjective Methods for Image Quality Assessment\" at Lodz University of Technology.\n\n## Acknowledgements\n\nThis project builds upon the NIMA implementation trained on the AVA dataset, as presented in Image Quality Assessment by Christopher Lennan, Hao Nguyen, and Dat Tran.\nPlease cite their work if applicable.\n\nGithub repository: https://github.com/idealo/image-quality-assessment\n\n## License\n\n- The code written by me is licensed under the [MIT License](licenses/LICENSE).\n- The following files were borrowed in an unchanged form from [Image Quality Assessment](https://github.com/idealo/image-quality-assessment) and are licensed under the [Apache 2.0 License](licenses/LICENSE-APACHE):\n  - vendor/nima_model_builder.py\n  - vendor/utils/losses.py\n  - vendor/weights_mobilenet_aesthetic_0.07.hdf5\n\n## Get Started\n\nFirst, clone this repository and install dependencies:\n```py\ngit clone https://github.com/mat-kubiak/iqa-thesis.git\ncd iqa-thesis\n\n# install dependencies in a virtual environment\npython3 -m venv env\nsource ./env/bin/activate.sh\npip install -r requirements.txt\n\n# or directly on your system\npip install tensorflow scikit-learn matplotlib flask opencv-python\n```\n\nAfterwards, images from CGIQA-6K should be placed in the `data/images/` directory.\nIn case of using another dataset of choice, the file `data/mos.csv` should also be replaced by one containing MOS scores of the new dataset.\n\n\u003e Unfortunately we cannot include those images here, as the whole dataset is 2.5GB in size. Please refer for it [here](https://github.com/zzc-1998/CGIQA6K).\n\nAfter providing all required data, run these scripts:\n\nscript in order to randomly split the dataset:\n```py\n# randomly split the data (required only once)\npython3 tools/split_db.py\n\n# you can customize the training script up to the `# USER PARAMS END HERE` comment\nvim train.py\n\n# train the network in background\nnohup python3 train.py\n\n# open tensorboard in the background for real-time monitoring\nnohup tensorboard --logdir=output --reload_interval=1 --window_title=\"Hybrid-CGIQA TensorBoard\" --port=6006 \u003e tensorboard-nohup.out 2\u003e\u00261 \u0026\n\n# test your network\nvim test.py\npython3 test.py\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmat-kubiak%2Fhybrid-cgiqa","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmat-kubiak%2Fhybrid-cgiqa","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmat-kubiak%2Fhybrid-cgiqa/lists"}