Fresh Class
Fresh Class🍎, A deep learning web application that classifies 6 types of fruits into 'Good' or 'Bad' quality using MobileNetV2 CNN. Features a Next.js frontend with real-time image classification. Team members: -Salah Eddine Berredjem -Ziad Abadlia -Tesnim Lala Bouali -Ines Goutel -Imen Kanoua -Mohamed Saber Tata
Summary
| Latest Version | Unknown |
|---|---|
| License | Unknown |
| CI Status | Failing |
| Stars | 1 |
| Forks | 0 |
| Open Issues | 0 |
| Last Commit | 2026-02-12 |
| Downloads | 0 |
| Last Indexed | 2026-09-07 06:08 |
Tags
Installation
nimble install Fresh Class
choosenim install Fresh Class
git clone https://gitlab.com/SalahBerr/fruit-quality-classifier
OS Compatibility
| Platform | Linux | macOS | Windows | FreeBSD | OpenBSD | NetBSD | Android | iOS | WASM | Embedded |
|---|---|---|---|---|---|---|---|---|---|---|
| Fresh Class | ✓ | ✓ | ✓ | - | - | - | - | - | - | - |
Source
| Repository | https://gitlab.com/SalahBerr/fruit-quality-classifier |
|---|---|
| Homepage | https://gitlab.com/SalahBerr/fruit-quality-classifier |
| Registry Source | gitlab |
README
FruitFresh AI - Fruit Quality Classification System 🍎
A deep learning web application that classifies 6 types of fruits into 'Good' or 'Bad' quality using MobileNetV2 CNN. Features a Next.js frontend with real-time image classification.
📋 Supported Fruits
- Apple 🍎
- Banana 🍌
- Orange 🍊
- Guava 🍏
- Lemon 🍋
- Pomegranate,
🌟 Features
- Multi-Fruit Classification: Supports 6 types of fruits (Apple, Banana, Orange, etc.)
- Quality Assessment: Classifies each fruit as 'Good' or 'Bad' quality
- Real-time Prediction: Instant classification with confidence scores
- User-friendly Interface: Clean, intuitive Next.js web interface
- Mobile Responsive: Works seamlessly on desktop and mobile devices
- High Accuracy: Powered by fine-tuned MobileNetV2 CNN model
📊 Dataset
- Total Images: 14,786
- Classes: 12 (Apple, Banana, Guava, Lime, Orange, Pomegranate × Good/Bad)
- Train/Val/Test Split: 70/15/15
📝 Development Timeline
- ✅ Data preprocessing (Week 1)
- ✅ Phase 1 training (Week 2)
- ✅ Phase 2 fine-tuning (Week 3)
- ✅ Model evaluation (Week 3)
- 🔄 Web interface (Week 4)
🧠 Model Details
- Base Model: MobileNetV2 (pretrained on ImageNet)
- Transfer Learning: Fine-tuned on custom fruit dataset
- Input Size: 128×128×3 RGB images
- Output: Binary classification (Good/Bad) for each fruit
- Accuracy: 99.41% on test set
🎯 Usage
- Upload Image: Select or drag-drop a fruit image
- Select Fruit Type: Choose from the 6 supported fruits
- Classify: Click "Analyze" button
- View Results: See prediction (Good/Bad) with confidence score
🌐 Deployment
This project is deployed on: - Frontend: Vercel - Backend: Hugging Face Spaces / Railway / Render - Model: MobileNet V2
Live Demo 🔗 [https://fruit-quality-classifier-jpyo.vercel.app/]
Prerequisites
- Python 3.8+
- Node.js 16+
- Git & Github
- TensorFlow 2+
- Google Colab Notebook
- Vercel
👥 Team
- Salah Eddine Berredjem - Group Leader
- Ines Goutel - Data preprocessing and cleaning
- Tesnim Lala Bouali - Model training and evaluation
- Imen Kanoua - Testing the Model
- Zied Abadliya - Web interface development
- Saber Tata - Documentation
