{"id":23105318,"url":"https://github.com/kartai/kai_mottakskontroll_objektdeteksjon","last_synced_at":"2026-07-05T19:31:48.265Z","repository":{"id":268462439,"uuid":"881851697","full_name":"kartAI/KAi_mottakskontroll_objektdeteksjon","owner":"kartAI","description":"Teste KartAI-algoritmer på utvalgt område i et GeoVekst-prosjekt, for å sjekke om KartAI-algoritmene kan detektere bygninger på et nivå som kan være til hjelp ved mottakskontroll av bygningsdata konstruert fra flybilder.","archived":false,"fork":false,"pushed_at":"2025-06-09T19:10:28.000Z","size":438,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-10-27T13:44:49.645Z","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":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/kartAI.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"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,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2024-11-01T11:29:56.000Z","updated_at":"2025-06-09T19:10:32.000Z","dependencies_parsed_at":"2024-12-16T23:34:10.206Z","dependency_job_id":"8fbe8498-fd13-4f47-97d0-c2d4d928ec25","html_url":"https://github.com/kartAI/KAi_mottakskontroll_objektdeteksjon","commit_stats":null,"previous_names":["kartai/kai_mottakskontroll_objektdeteksjon"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/kartAI/KAi_mottakskontroll_objektdeteksjon","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kartAI%2FKAi_mottakskontroll_objektdeteksjon","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kartAI%2FKAi_mottakskontroll_objektdeteksjon/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kartAI%2FKAi_mottakskontroll_objektdeteksjon/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kartAI%2FKAi_mottakskontroll_objektdeteksjon/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/kartAI","download_url":"https://codeload.github.com/kartAI/KAi_mottakskontroll_objektdeteksjon/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kartAI%2FKAi_mottakskontroll_objektdeteksjon/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35167259,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-07-05T02:00:06.290Z","response_time":100,"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":[],"created_at":"2024-12-17T00:49:53.521Z","updated_at":"2026-07-05T19:31:48.244Z","avatar_url":"https://github.com/kartAI.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Deep Learning for Building Detection for Completeness Check in Mottakskontroll – Master Thesis\n\nThis repository contains the code developed as part of a master's thesis at NTNU. The project investigates how deep learning can support the Norwegian Mapping Authority’s *mottakskontroll* (completeness check) for FKB-building data by detecting buildings in aerial imagery using instance segmentation models.\n\n## Project Structure\n\nAll the code produces in the master's thesis is in the master folder. The master folder have the following folders: \n\n - mask_r_cnn: All code related to Mask R-CNN using Detectron2\n- yolo: All code related to YOLOv8-seg using Ultralytics\n- preprocessing: Scripts to tile TIFFs and generate training data\n- scripts: Miscellaneous utilities (visualization, evaluation, etc.)\n\n## Models and Frameworks\n\nThe following models and tools are used in the project:\n\n- **YOLOv8-seg** [Ultralytics](https://github.com/ultralytics/ultralytics) \n- **Mask R-CNN** [Detectron2](https://github.com/facebookresearch/detectron2)\n- **SAHI** – [Sliced Aided Hyper Inference(SAHI)](https://github.com/obss/sahi) to improve predictions near tile edges \n\n## Installation\n\nThis project uses Python 3.12+. Dependencies are declared in the `pyproject.toml` file. Minimal installation example:\n\n```toml\n[project]\nname = \"master\"\nversion = \"0.1.0\"\ndescription = \"Deep learning models for building detection in aerial images\"\nreadme = \"README.md\"\nrequires-python = \"\u003e=3.12\"\ndependencies = [\n    \"torch\u003e=2.6.0\",\n    \"torchvision\u003e=0.21.0\",\n]\n```\nTo install dependencies, use a tool like pip or poetry:\n```\npip install torch torchvision\n````\n\n⚠️ Additional installation steps are required for Ultralytics and Detectron2. Follow their official guides based on your OS and CUDA version.\n\n## Data Requirements\nTo run the full pipeline, you need:\n\n- Georeferenced TIFF images\n- FKB building data in polygon format stored as a GeoPackage (.gpkg)\n\nThese are processed into training tiles and annotations using the preprocessing/ scripts. Both COCO and YOLO format datasets are supported.\n\n## Usage\nPreprocess the data\nUse the scripts in preprocessing/ to tile the input imagery and convert annotations to COCO or YOLO formats.\n\nTrain models\n\nRun training scripts in yolo/ for YOLOv8-seg\n\nRun training scripts in mask_r_cnn/ for Mask R-CNN (Detectron2)\n\n\n## Acknowledgements\nThis project was conducted as part of a master’s thesis in Engineering and ICT at NTNU, in collaboration with the Norwegian Mapping Authority (Kartverket) and the KartAI initiative.\n\n## Note\nIf you use this code or build upon it, please cite the frameworks used:\n\nUltralytics YOLOv8\nDetectron2\nSAHI\n\nThis repository is intended for academic and research use. \n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkartai%2Fkai_mottakskontroll_objektdeteksjon","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkartai%2Fkai_mottakskontroll_objektdeteksjon","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkartai%2Fkai_mottakskontroll_objektdeteksjon/lists"}