{"id":31681346,"url":"https://github.com/romanslack/refacenet","last_synced_at":"2025-10-08T07:48:23.781Z","repository":{"id":295634716,"uuid":"990660192","full_name":"RomanSlack/ReFaceNet","owner":"RomanSlack","description":"ReFaceNet intelligently reconstructs complete faces from multiple occluded/cropped photos. Handles hand-covered selfies, partial crops, and missing facial regions. Features smart    alignment, multi-scale blending, and coverage heatmaps. Perfect for completing partial face datasets.","archived":false,"fork":false,"pushed_at":"2025-05-26T15:31:05.000Z","size":73204,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-05-26T15:58:28.485Z","etag":null,"topics":["ai","image-processing"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/RomanSlack.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}},"created_at":"2025-05-26T12:48:16.000Z","updated_at":"2025-05-26T15:31:50.000Z","dependencies_parsed_at":"2025-05-26T16:08:54.916Z","dependency_job_id":null,"html_url":"https://github.com/RomanSlack/ReFaceNet","commit_stats":null,"previous_names":["romanslack/refacenet"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/RomanSlack/ReFaceNet","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/RomanSlack%2FReFaceNet","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/RomanSlack%2FReFaceNet/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/RomanSlack%2FReFaceNet/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/RomanSlack%2FReFaceNet/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/RomanSlack","download_url":"https://codeload.github.com/RomanSlack/ReFaceNet/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/RomanSlack%2FReFaceNet/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":278909712,"owners_count":26066887,"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":"2025-10-08T02:00:06.501Z","response_time":56,"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":["ai","image-processing"],"created_at":"2025-10-08T07:48:03.331Z","updated_at":"2025-10-08T07:48:23.772Z","avatar_url":"https://github.com/RomanSlack.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# ReFaceNet\n\n![Latest Reconstruction](outputs/latest_reconstruction.png)\n\n![Latest Coverage Heatmap](outputs/latest_heatmap.png)\n\nReFaceNet reconstructs a complete 2D face from multiple occluded or cropped images of the same person. It intelligently combines visible face parts from different images to create a unified facial reconstruction, handling partial occlusions, hand-covered faces, and cropped photos.\n\n## How It Works\n- **Smart Face Detection**: Detects faces even when partially occluded by hands or objects\n- **Standardized Alignment**: Aligns all faces to a common template for consistent geometry\n- **Multi-Scale Blending**: Combines visible regions at different detail levels for natural results\n- **Coverage Analysis**: Tracks which facial regions are reconstructed vs missing\n- **Quality Weighting**: Prioritizes clearer, higher-quality facial regions\n\n## Key Features\n- **2D Face Reconstruction**: Complete facial images from partial inputs\n- **Occlusion Handling**: Works with hand-covered selfies and cropped photos\n- **Multi-method landmark detection** with robust fallbacks\n- **Feature-specific face masking** (eyes, nose, mouth, cheeks, forehead)\n- **Real-time coverage heatmaps** showing reconstruction quality\n- **Generation tracking** for iterative improvements\n\n## Quick Start\n\n### 1. Setup Environment\n```bash\n# Follow setup instructions in SETUP_GUIDE.md\nconda create -y -n ReFaceNet python=3.10\nconda activate ReFaceNet\npip install opencv-python mediapipe tqdm\n```\n\n### 2. Add Your Images\nPut your face images (can be partially occluded) in `data/raw/`\n\n### 3. Run Reconstruction\n```bash\npython run_pipeline_smart.py\n```\n\n### 4. View Results\n- `outputs/latest_reconstruction.png` - Your reconstructed face\n- `outputs/debug/` - Debug visualizations\n- `face_reconstruction.log` - Detailed processing log\n\n## Output Files\n- `generation_X_reconstruction.png` - Numbered results for tracking progress\n- `latest_reconstruction.png` - Always the most recent result\n- `generation_X_heatmap.png` - Numbered coverage heatmaps showing which face regions were reconstructed\n- `latest_heatmap.png` - Always the most recent coverage heatmap\n- `generation_X_stats.json` - Processing statistics for each run\n- `debug/` folder - Individual aligned images and landmark visualizations\n\n## Understanding the Heatmap\nThe coverage heatmap shows which parts of the face were successfully reconstructed:\n- **Red/Yellow**: High coverage - multiple input images contributed to this region\n- **Green/Blue**: Medium coverage - some input images contributed  \n- **Dark Blue/Black**: Low/no coverage - few or no input images had visible data for this region\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fromanslack%2Frefacenet","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fromanslack%2Frefacenet","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fromanslack%2Frefacenet/lists"}