SwipeClearn

Allow users to clean up their photo library based on when the photo was added (e.g., focus cleanup only on the last week, month, or year). This gives users tighter control and a feeling of quick wins.

Comment

  • Inspiration Every smartphone user has thousands of photos—many of them blurry, duplicate, or screenshots long forgotten. Deleting them is tedious, and most people don’t do it until they run out of space. We wanted to create a fun, mobile-first tool that makes it effortless to declutter your digital gallery while also making users aware of the hidden environmental cost of data storage.

  • What it does SwipeClean helps users clean up their photo library using a swipe interface powered by AI. It: Detects low-quality, duplicate, or irrelevant photos Lets users swipe left to delete or right to keep Allows filtering by album and timeframe (e.g., last 7 days) Estimates storage space freed and CO₂ impact reduced Gives users a quick impact summary at the end of their session

  • How we built it Frontend: Built in React Native with Expo for a smooth mobile experience Photo Access: Used Expo MediaLibrary API to fetch and filter albums and assets AI Classification: Integrated TensorFlow.js MobileNet to classify photos by type (blurry, screenshot, duplicate) Swipe Interface: Custom SwipeDeck component using gesture responders CO₂ Impact: Estimated carbon savings using storage size + cloud energy assumptions Timeframe Filter: Implemented TimeframeSelector.js and logic to filter by recent days

  • Challenges we ran into Accessing and filtering media across Android and iOS with consistent performance Building a real-time AI pipeline that runs on-device without major lag Creating a swipeable UX that feels natural and intuitive Estimating environmental impact in a way that is meaningful but lightweight Managing performance with large photo libraries

  • Accomplishments that we're proud of Fully working prototype in under 48 hours with real swipe interactions and AI classification Clean, mobile-optimized UI that feels familiar and effective Environmental impact estimation that gives real value beyond storage cleanup Modular architecture for fast extension into document and email decluttering

  • What we learned TensorFlow.js can be surprisingly performant on-device when used properly Expo’s MediaLibrary is powerful but tricky when dealing with permissions and large datasets Simple design patterns (like swiping) can dramatically improve user engagement CO₂ savings from deleting digital clutter is a real and underexplored topic Time-based filters give users a sense of momentum and progress

  • What's next for SwipeClean V2: Document decluttering: PDF/doc classification and swipe cleanup V3: Email decluttering: Connect Gmail to clean newsletters, spam, old threads Add cloud photo sync integrations (iCloud, Google Photos) Let users track cumulative CO₂ saved over time (impact profile) Introduce gamification: streaks, badges, cleanup challenges Publish to iOS + Android app stores

Built With

Share this project:

Updates