Android WhatsApp Property Extractor App

I receive a constant stream of property listings through WhatsApp groups and direct messages. I need an Android application that can quietly sit in the background, read every incoming message or image, decide whether it is junk, and—when it is relevant—pull out the key facts and save them to a searchable, well-structured database. Core workflow • Capture all incoming WhatsApp notifications (text or image) in real time. • Classify each message as “junk” or “property lead.” I’m open to your recommendation here—keyword rules, a lightweight ML model, or any hybrid you feel delivers the best accuracy without draining battery. • For leads, extract: – Location (break it into up to multiple granular columns) – Size, price, magnetic orientation, special features, sender details and any other typical listing data you can detect • When the listing arrives as an image, run on-device OCR (e.g., Google ML Kit, Tesseract or equivalent) before extraction. • Store everything in an easy-to-query local database(I'm open to recommendations here on what kind of database would be best for this use case) with a simple in-app table view and search bar so I can quickly filter by price, area, orientation, etc. Nice-to-have touches – A one-tap “export to CSV” or Google Sheets sync – A confidence score for each extracted field so I know when to double-check the data Acceptance criteria 1. App captures 100 % of incoming WhatsApp messages and images while device is unlocked or unlocked. 2. At least 98% of clearly formatted listings are identified as non-junk. 3. Extracted fields populate correctly into the database and are visible in the UI search. 4. APK runs on Android 10+. Let me know which classification approach you recommend and any libraries you plan to use for OCR, NLP, and database handling. I’m ready to start as soon as you can outline your proposed tech stack and timeline.

Project ID: 40189425