About the project
Fresh fruits and vegetables in Indian mandis are still graded by eye. It is slow, every trader judges differently, farmers get one flat price even for a good lot, and hotels or retailers often receive lower quality than promised. Fruit and vegetable quality grading Android project FreshGrade solves this with the phone camera. The seller takes 1 to 5 photos of a lot, a CNN image classification model running on the phone gives Fresh, Average and Spoiled percentages, and the app turns them into a clear Grade A, B or C with a suggested price, shelf-life estimate and a PDF quality certificate. Buyers see the grade before they buy and can re-check the same produce on delivery.
How it works
- Seller selects the produce (tomato, onion, mango and 13 more) and takes 1-5 photos with the camera or gallery
- MobileNetV2 CNN in TensorFlow Lite grades every photo on the phone: score = fresh% + 0.55 x average% + 0.10 x spoiled%
- Batch score decides the grade: A (Fresh) 80+, B (Average) 50+, else C (Spoiled) - limits set by admin
- Suggested price = today's mandi rate x grade factor (A 1.10, B 0.85, C 0.40) plus shelf life in days
- Seller publishes a listing with the grade badge and quality certificate; Grade C cannot be sold
- Buyer orders, seller accepts, dispatches and delivers; stock is reserved automatically
- Buyer re-checks quality on delivery; if the grade dropped, the app calculates the fair price and claim, and the admin resolves the dispute
Users
- Seller (farmer / vendor / FPO): scan and grade batches, price suggestion, listings, orders, disputes, quality reports, trust badge
- Buyer (hotel / retailer / mess): shop graded produce, cart and checkout, order tracking, delivery re-check, disputes, ratings, purchase summary
- Admin (web panel): approve sellers, produce and mandi rate masters, graded batches log, orders, disputes, notices, grading settings, printable reports
Highlights
- 3 roles, about 25 Android screens, 16 admin web pages, 29 PHP API files, 14 database tables
- On-device CNN (MobileNetV2 transfer learning) with an offline colour-analysis fallback, so the app works even before the model is trained
- Ready Google Colab notebook to train the model on free Kaggle datasets (FruitNet + Fresh/Stale fruits and vegetables)
- PDF quality certificate and order invoice made in the app
- Seller trust score with Gold / Silver / Bronze badges from ratings and re-check results
- Charts: seller quality trend and grade split, buyer weekly spend, admin grade chart
- Validations on phone and server: email, 10-digit mobile, pincode, quantity and price limits, date pickers, photo type and size, re-check time window
- Secure PHP: prepared statements, password hashing, form tokens in the admin panel
- Sample data with demo logins for seller, buyer and admin
What you get
- Complete Android Studio source code (Java) and PHP admin panel + APIs
- MySQL database file with sample data
- Model training script and Colab notebook
- Project report (Word, 60 pages) with all UML diagrams, DFD and ER diagram
- Project abstract (Word + PDF), project explanation with viva questions
- Step-by-step setup guide (PDF) and installation help
Who can use this project
Final year students of BE / B.Tech Computer Engineering, IT and AI & DS, MCA, BCA and Diploma (Computer / IT) looking for an Android project with source code, a machine learning image classification project, a TFLite CNN Android project or an agriculture-based final year project with documentation and report.
Frequently asked questions
Does the app need internet to grade a photo? No. Grading runs on the phone with TensorFlow Lite; internet is needed only to save the batch and for listings and orders.
Is the trained model included? A ready Colab notebook trains it in about 30-40 minutes on a free GPU. Until then the app grades with a built-in colour check, so all demos still work.
Can I run it on a real phone? Yes. Enter your laptop's Wi-Fi IP on the login screen; the emulator uses 10.0.2.2.
Which produce items are supported? 16 items such as tomato, onion, potato, capsicum, mango, banana, apple and pomegranate; the admin can add more.
