Project Info
Inspiration
Users often are using phones to do photos to store memories from holiday, photos of friends, record price in the shop, receipt for warranty, item for sell, letter or document for further proof etc. Some of them are blurry other are duplicates. Image folders quietly become a mix of everything. Cloud photo solution like OneDrive and Google Photos are creating collections which works great but the mess on the folder location stay this same and is taking space which could be released. I wanted to have a practical way to deal with it with privacy in mind - without uploading private images to a cloud service or trusting an opaque automatic cleanup sites.
What it does
Clearframe is a local-first Windows desktop app for reviewing an image folder safely. It scans only folders the user selects, then explains its recommendations for advised category like duplicates, screenshots, documents, thumbnails, low-quality images, photos, and items needing review. Nothing is moved, deleted, or edited during scanning and user can decide what should happen with it next.
How we built it
My set of mind has been changed from "it's to complex, it's to time consuming" to "with AI it's possible and doable". Codex accelerated the project from an empty workspace through the desktop workflow, test suite, demo data, performance investigations, offline OCR packaging, and calibration features. I directed the product and safety decisions, reviewed the implementation, and repeatedly verified the result with automated tests and linting. This project has been build with: Python 3.12 desktop application with PySide6/Qt. Pillow, NumPy, and OpenCV provide local image analysis Pytesseract and a bundled Tesseract runtime provide optional offline OCR. Worker threads keep scanning and OCR off the UI thread. PyInstaller and NSIS package the app as a per-user Windows installer with no API key, administrator permission, or download required.
Challenges we ran into
The hardest challenge was to create/balance meaning full decision process about automatic category selection for the photo. My goal was to give user control over how the "decision" about category is made (user have tool which will based on the sample data suggest improvements to make the categorization work better). Performance: I found that full-resolution analysis could make scans feel slow. Bounded previews, cached local detectors, and an opt-in Deep Scan kept the everyday workflow responsive while preserving privacy. Clear Understanding: I learned to make recommendations explainable, keep actions approval-gated, and allow users quickly manually change it.
What's next
At some point I would like to introduce more advance OCR/recognition objective. Users still receiving letters, making photos of important documents, agreements, receipts. This can be easily lost and keep disorganized. My goal would be automatically sort it and categorized based on content so it can be searchable and user can use it in future instead of real paper as proof.
This project did not link a GitHub repository.
Analysis
No indexed repository for this project, so there are no commit stats to show.
Technology
- PythonUnchecked
No repository was indexed for this project, so these Devpost claims have not been checked against code.
AI coding agents
No repository was indexed, so agent usage could not be checked.
Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.
Codebase size
No repository was indexed, so there is no codebase to measure.
This project did not link a GitHub repository, so there is nothing to diagram.
This project did not link a GitHub repository, so its feature claims have not been checked against code.
Export this project's context (description, README, evidence, key source files) to chat with an AI agent elsewhere.