QUESTNET // PROJECT CHRONICLEFILE 3070

Turn a Clover Hunt into a Computer Vision Adventure
Develop BrentWare: CloverVision, a free BrentCorp Android application that uses a smartphone camera and custom computer vision to help people spot naturally occurring four-leaf clovers. The app should be fun, accessible, privacy-first and useful without internet access.
Core Features
- Display a live phone-camera view with clear visual markers around promising four-leaf clover candidates.
- Run image analysis on-device without mandatory accounts, cloud processing or subscriptions.
- Let users capture and save annotated photographs of candidate discoveries.
- Show an honest confidence indicator and prompt users to verify that four leaflets belong to one stem.
- Offer a simple guide to distinguishing four-leaf clovers from overlapping ordinary clovers.
Development Milestones
- Collect permissioned photographs of clover patches in varied lighting and leaf overlap conditions.
- Annotate true four-leaf examples, three-leaf examples and confusing lookalikes.
- Prototype candidate detection on still photographs and measure precision and recall.
- Optimize a lightweight model for offline Android inference and a responsive camera overlay.
- Test on real lawns, measure false positives and battery use, and refine the interface.
- Release the app for free with clear privacy documentation and a feedback channel.
Victory Condition
Publish a functional, freely available Android app that can scan live camera images offline, highlight potential four-leaf clovers, and save annotated photos, with documented real-world test results and limitations.
Stretch Goals
Support five-leaf detection, saved scan grids, accessibility options, open-source model/data tooling where licensing and privacy allow, and a future iOS version.
Related Quest
Find a Four-Leaf Clover. CloverVision is an optional tool for that adventure, not a requirement.
