Project Files • (fruits_quality_system.py )AI model training, prediction functions, and image processing • (streamlit_app.py) Interactive web dashboard with real-time analysis • (produce_fruit_veg_quality.py )SQLite database storing all predictions and analytics • (improved_produce_quality_model.h5 )Trained TensorFlow model (76.67% accuracy)

Model Performance • Classification Accuracy: 76.67% • Freshness MAE: 1.16 • Quality Classes: 4 (Unripe, Ripe, Overripe, Bruised) • Validation Performance: Comprehensive testing completed

Testing & Validation Accuracy Metrics • Target Accuracy: 90% (Industry standard) • Achieved Accuracy: 76.67% (Solid baseline) • Validation Method: Train/Test split with cross-validation

Technical Features Image Processing • Background removal using HSV thresholding • Multi-color space analysis (RGB, HSV, LAB) • Texture feature extraction • ROI detection and normalization AI/ML Capabilities • Transfer Learning with MobileNetV2 • Custom multi-output architecture • Early stopping and learning rate scheduling • Comprehensive model evaluation Web Interface • Real-time webcam integration • Interactive visualizations • Responsive design • Database connectivity