Project Introduction
The "Design and Implementation of a Mud Crab Gender Classification and Counting System" is an undergraduate thesis project developed to automate the identification and counting of mud crabs using machine learning technology. The system aims to assist aquaculture farmers and researchers by providing a faster, more accurate, and efficient alternative to manual mud crab classification.
About the Project
This project utilizes image processing and machine learning techniques to detect mud crabs, determine their gender, and automatically count the total number present in an image or camera feed. The system was developed using Python and integrated with a trained classification model capable of distinguishing male and female mud crabs based on their abdominal flap.
Key Features
- Automated mud crab detection and counting
- Gender classification (Male/Female)
- Real-time image processing capability
- User-friendly graphical interface
- Improved accuracy and efficiency compared to manual counting methods
Technologies Used
- Python
- OpenCV
- Machine Learning
- YOLO Object Detection (YOLOv11)
- Tkinter GUI
- Raspberry Pi Camera Module
Step-by-Step Process
1. Collection of Data
Collect a diverse dataset of images containing mud crabs in various environments and lighting conditions.
2. Annotation of Data
Label the collected images with polygon tool around the abdominal flaps of the mud crabs and their corresponding labels.
3. Machine Learning Model Training (YOLOv11)
Train the YOLOv11 model on the annotated dataset to enable accurate detection and classification of mud crabs.
4. Implement Model to the Python Application
Integrate the trained YOLOv11 model into a Python application, ensuring seamless functionality and user-friendly interface.
Screenshots
Results
The developed system successfully identified and counted mud crabs while accurately classifying their gender. Testing demonstrated that the system could process images efficiently and reduce the time required for manual classification. The project highlights the potential application of artificial intelligence and computer vision in modern aquaculture management and monitoring.
Technical Write-Up
For a deeper look at the hardest part of this project — training a model to reliably tell apart subtle abdominal flap differences on a small, imbalanced dataset — read the full technical breakdown.
Read the Full Write-Up on dev.toGitHub Repository
This repository contains the source code for the Mud Crab Gender Classification and Counting System. It serves as the primary source for project development, version control, and future enhancements.
View Repository