Tilapia Fish Freshness Evaluation System

A computer vision system for evaluating tilapia fish freshness using YOLOv3 object detection, GrabCut segmentation, and advanced RGB color analysis.

Features

• Object Detection: Custom YOLOv3 model for tilapia gill identification
• Segmentation: GrabCut algorithm for precise gill area extraction
• Color Analysis: Advanced RGB analysis with brightness, saturation, and color ratios
• GUI Interface: User-friendly Tkinter-based interface with progress indicators
• Performance Optimizations: Image caching, lazy loading, and memory-based processing
• Structured Logging: Industry-standard logging with separate analysis results
• Modern Architecture: Object-oriented design with proper separation of concerns

Project Structure

├── src/tilapia_freshness/
│   ├── models/              # ML models (YOLO, GrabCut, Analyzer)
│   ├── gui/                 # GUI components
│   ├── utils/               # Utility functions (caching, progress, logging)
│   ├── enums/               # Enums and constants
│   ├── config/              # Configuration management
│   ├── exceptions/          # Custom exceptions
│   └── assets/              # Application resources (icon)
├── tests/                   # Unit tests
├── .github/workflows/       # CI/CD pipeline
├── images/                  # Sample images
├── inputs/                  # Standardized input files (auto-generated)
├── outputs/                 # Generated results and logs
├── main.py                  # Application entry point
├── pyproject.toml           # Modern Python packaging
└── requirements.txt         # Dependencies

Prerequisites

• Python 3.11+
• Required packages listed in requirements.txt

Installation

1. Clone Repository

git clone https://github.com/Evrouin/tilapia-fish-freshness-evaluation-system.git
cd tilapia-fish-freshness-evaluation-system

2. Setup Virtual Environment

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install Dependencies

pip install -e .
# Or for development:
pip install -e ".[dev]"

4. Download Model Files

Download the required model files from Google Drive and place them in the data/ folder:

• yolov3_tilapia.weights
• yolov3_tilapia.cfg
• obj.names

Usage

Run Application

python main.py
# Or if installed:
tilapia-freshness

Using the GUI

1. Click "Select Image" to choose a tilapia fish image
2. The system automatically detects gills using YOLOv3 with progress indication
3. Click "Start Analysis" to perform segmentation and freshness evaluation
4. Results display RGB values, confidence scores, and freshness classification

Freshness Classifications

• Fresh: Good color quality, red dominance, appropriate saturation
• Not Fresh: Moderate degradation, unclear color characteristics
• Old: Blue/gray dominance, washed out colors, high brightness

Development

Development Setup

make install-dev    # Install with development dependencies

Code Quality

make format         # Format code with Black and isort
make lint          # Run flake8 linter
make type-check    # Run mypy type checker
make test          # Run tests with coverage
make check-all     # Run all quality checks

Running Tests

pytest tests/ -v --cov=src/tilapia_freshness

Technical Details

Key Components

• YOLODetector: Handles gill detection with lazy loading
• GillSegmenter: Performs GrabCut segmentation
• ColorAnalyzer: Analyzes RGB values and classifies freshness
• InputFileManager: Standardized file processing with timestamp naming
• ProgressDialog: User feedback for long operations

Color Analysis Algorithm

The system uses multiple metrics for accurate freshness evaluation:

• Distance from fresh reference point (32, 19, 17)
• Color brightness and saturation analysis
• RGB ratios and dominance patterns
• Weighted mean/median filtering for noise reduction
• Configurable thresholds for different freshness levels

Authors

• @John Elway Cortez - Lead Developer
• @Dustin Uriel Obispo - Co-Developer
• @King Red Sanchez - Co-Developer
• @Briel Aldous Viola - Co-Developer

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

Please read CONTRIBUTING.md for details on our code of conduct and the process for submitting pull requests.