{"id":27648200,"url":"https://github.com/pathwiselabs/pixel-pipeline","last_synced_at":"2026-03-04T22:31:53.025Z","repository":{"id":289148474,"uuid":"967933638","full_name":"pathwiselabs/pixel-pipeline","owner":"pathwiselabs","description":"A Python application with Gradio UI for batch processing and captioning of images, allowing for easy integration with AI image training workflows.","archived":false,"fork":false,"pushed_at":"2025-05-20T20:32:14.000Z","size":413,"stargazers_count":7,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-06-15T22:40:19.075Z","etag":null,"topics":["data-cleaning","data-science","flux","generative-ai","stable-diffusion","stable-diffusion-webui"],"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/pathwiselabs.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,"zenodo":null}},"created_at":"2025-04-17T08:25:19.000Z","updated_at":"2025-06-01T05:23:29.000Z","dependencies_parsed_at":"2025-06-15T22:36:04.679Z","dependency_job_id":"4b0d52cf-8d8d-4774-8065-144f131627e5","html_url":"https://github.com/pathwiselabs/pixel-pipeline","commit_stats":null,"previous_names":["pathwiselabs/pixel-pipeline"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/pathwiselabs/pixel-pipeline","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/pathwiselabs%2Fpixel-pipeline","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/pathwiselabs%2Fpixel-pipeline/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/pathwiselabs%2Fpixel-pipeline/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/pathwiselabs%2Fpixel-pipeline/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/pathwiselabs","download_url":"https://codeload.github.com/pathwiselabs/pixel-pipeline/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/pathwiselabs%2Fpixel-pipeline/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30096762,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-04T21:59:23.547Z","status":"ssl_error","status_checked_at":"2026-03-04T21:57:50.415Z","response_time":59,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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":["data-cleaning","data-science","flux","generative-ai","stable-diffusion","stable-diffusion-webui"],"created_at":"2025-04-24T02:35:01.525Z","updated_at":"2026-03-04T22:31:53.017Z","avatar_url":"https://github.com/pathwiselabs.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cimg src=\"resources/readme_header.png\" alt=\"Pixel Pipeline Logo\" width=\"500\"/\u003e\n\n## What is it?\n\nPixel Pipeline is a desktop application that helps you refine image datasets for AI training by cleaning, filtering, and enhancing your image collections. It offers a comprehensive set of tools for removing duplicates, filtering images by face count, reducing dataset size while maintaining diversity, and generating high-quality AI captions.\n\n## Features\n\n### Complete Image Dataset Refinement Workflow\n\n- **Image Similarity Detection**: Remove duplicates and visually similar images\n  - Perceptual hash for exact duplicates\n  - VGG16-based similarity detection for visually similar images\n\n- **Face Detection \u0026 Filtering**: Sort images by face count\n  - Multiple detection algorithms (MTCNN, MediaPipe, YuNet, etc.)\n  - Automatic sorting into no-face, single-face, and multiple-face categories\n\n- **Image Set Refinement**: Intelligently reduce dataset size\n  - Face embedding and clustering\n  - Diversity or consistency-based selection\n  - Interactive visualization of clustering results\n\n- **AI-Powered Image Captioning**: Generate high-quality captions\n  - Qwen2.5-VL vision-language models (3B \u0026 7B variants)\n  - Custom prompting for tailored descriptions\n  - Flux-compatible identifier generation\n  - Caption export as ZIP files\n\n## System Requirements\n\n- **Python**: 3.11.9 or compatible\n- **CUDA**: 12.8 (Blackwell compatible)\n- **PyTorch**: Nightly builds of torch and torchvision (Blackwell compatible)\n- **RAM**: 8GB minimum, 16GB+ recommended\n- **GPU**: NVIDIA GPU with 6GB+ VRAM recommended for optimal performance\n- **Storage**: 10GB+ free space for models and temporary files\n\n## Installation\n\n1. Clone this repository:\n```bash\ngit clone https://github.com/yourusername/pixel-pipeline.git\ncd pixel-pipeline\n```\n\n2. Create and activate a virtual environment:\n```bash\npython -m venv venv\n\n# On Windows\nvenv\\Scripts\\activate\n\n# On macOS/Linux\nsource venv/bin/activate\n```\n\n3. Install the required dependencies:\n```bash\npip install -r requirements.txt\n```\n\n**Note for CUDA compatibility**: This project requires CUDA 12.8 compatible PyTorch. The requirements.txt file includes the correct version specifications.\n\n## Usage\n\n### Launching the Application\n\nRun the application using the included batch file (recommended):\n\n```bash\nrun.bat\n```\n\nOr launch directly with Python:\n\n```bash\npython app.py\n```\n\nThe application will open in your default web browser at http://localhost:7865.\n\n### Recommended Workflow\n\nFor best results, follow this sequence:\n\n1. **Image Similarity Tab**: First remove duplicates and overly similar images\n2. **Face Detection Tab**: Next, filter images based on face count (for character LoRA training)\n3. **Image Set Refinement Tab**: Reduce dataset size while maintaining diversity or consistency\n4. **Image Captioning Tab**: Finally, add AI-generated captions to your refined dataset\n\n## FAQ\n\n### Why am I getting CUDA/GPU errors?\nEnsure you have NVIDIA drivers installed that support CUDA 12.8. You can also try using a smaller batch size or the 3B model variant for captioning if you have limited VRAM.\n\n### Can I run this without a GPU?\nYes, but performance will be significantly slower, especially for face detection, clustering, and captioning tasks.\n\n### How do I report a bug or request a feature?\nPlease open an issue on the GitHub repository with a detailed description of the bug or feature request.\n\n### What image formats are supported?\nPixel Pipeline supports JPG, PNG, and JPEG image formats.\n\n### How many images can I process at once?\nThis depends on your hardware. With sufficient RAM and VRAM, you can process hundreds or thousands of images, but for optimal performance, consider batching large datasets.\n\n## Contributing\n\nContributions are welcome! Please feel free to submit a Pull Request.\n\n1. Fork the repository\n2. Create your feature branch (`git checkout -b feature/amazing-feature`)\n3. Commit your changes (`git commit -m 'Add some amazing feature'`)\n4. Push to the branch (`git push origin feature/amazing-feature`)\n5. Open a Pull Request\n\n## License\n\nMIT License\n\n## Acknowledgments\n\n- This application utilizes the [Qwen2.5-VL](https://github.com/QwenLM/Qwen-VL) model from Alibaba\n- Built with [Gradio](https://www.gradio.app/) for the user interface\n- [Hugging Face Transformers](https://huggingface.co/docs/transformers/index) for model integration\n- Face detection and analysis powered by various open-source libraries including DeepFace and FaceNet\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fpathwiselabs%2Fpixel-pipeline","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fpathwiselabs%2Fpixel-pipeline","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fpathwiselabs%2Fpixel-pipeline/lists"}