CHAN SOPHARA

A chatbot assistant integrated into a food delivery system to handle customer inquiries and orders with an interactive and modern design.

Project Details

• Added chatbot assistant feature to existing food delivery system
• Created separate server using Flask specifically for the chatbot functionality
• Implemented handling for orders, menu checking, location services, special offers, social media links, and team information
• Designed conversational flows for natural user interaction
• Integrated with the main food delivery system's database

Technologies Used

A complete food delivery platform for Khmer restaurants with customer interface, admin dashboard, and order management system.

Project Details

• Designed user-friendly and interactive Front End using HTML, CSS, and JavaScript
• Developed Back End using Flask server with SQLite and MySQL databases
• Created separate Admin page and dashboard for real-time customer data monitoring
• Implemented features for handling customer orders and printing order transcripts
• Optimized for both desktop and mobile viewing

Technologies Used

A full-stack web application for managing gym memberships, staff, and facility bookings with real-time data processing.

Project Details

• Developed responsive frontend using HTML, CSS, and JavaScript
• Built RESTful API backend with Node.js and Express
• Implemented CRUD operations for member and staff management
• Designed relational database schema and deployed with XAMPP (PHPMyAdmin)
• Created user authentication and session management
• Integrated real-time data updates without page refresh

Key Features

• Member registration and profile management
• Class scheduling and booking system
• Attendance tracking and reporting
• Payment processing integration
• Responsive design for all device sizes

Technologies Used

This project aims to reduce accidents caused by drowsy driving by detecting signs of drowsiness in real-time and alerting the driver.

Project Details

• Built a machine learning model to detect drowsiness using facial landmarks and eye aspect ratio
• Implemented two models: ZFNet (89.59% accuracy) as primary and CNN (93.96% accuracy) as fallback
• Connected with Arduino to provide physical alerts (vibration and sound) when drowsiness is detected
• Optimized for real-time performance with OpenCV for video processing

Technologies Used

A comprehensive system to detect and classify traffic lights and signs in real-time, suitable for autonomous vehicles and driver assistance systems.

Project Details

• Built a model to detect traffic lights and signs using YOLOv8 with 99% image accuracy and 80% video accuracy
• Implemented classification using CNN with 97.84% image accuracy
• Optimized for real-time performance with frame processing at 30 FPS
• Trained on custom dataset of Cambodian traffic signs and lights

Technologies Used

A machine learning application that predicts loan default risk based on customer data, deployed as a web application.

Project Details

• Collected and preprocessed loan data from Kaggle
• Applied various machine learning algorithms to build prediction models
• Performed data analysis and visualization to understand feature importance
• Deployed the best performing model as a web application
• Created interactive visualizations to explain model decisions

Technologies Used