All projects
Year 3 · Academic Year 2025

Year 3 Projects

DMT321 Integrated Product Design and Development / DMT322 Digital Image Processing Technique / MMI320 Medical Instruction Media Technology / MDT312 Web Programming

Academic Year 2025

IoT, AR, Computer Vision & Web Development

DMT321 Integrated Product Design and Development / DMT322 Digital Image Processing Technique / MMI320 Medical Instruction Media Technology / MDT312 Web Programming

  • Designed and built a smart walking stick for elderly users on an ESP32 with motion (MPU6050) and grip-pressure (FSR402) sensors, fall detection and Wi-Fi app alerts.
  • Built a Unity digital twin of a real junction where YOLO object detection counts cars from virtual CCTV and adapts the traffic lights, cutting peak-hour delay by 58%.
  • Built HemoLens AR, a marker-based AR app in Unity and Vuforia that teaches blood-group compatibility with 3D blood-cell models.
  • Built my first portfolio website and UnNiYam Shop, a full-stack online store with Node.js, Express and MySQL.
ESP32 IoT Unity AR YOLO Computer Vision JavaScript Node.js MySQL

Course Projects

See each project and open its files, live site or demo.

DMT321

Integrated Product Design and Development

ESP32 MPU6050 FSR402 OLED Wi-Fi
Smart Walking Stick prototype: the stick with its sensor handle, wiring and the laptop dashboard
IoT Product · Group project (3 members)

Smart Walking Stick

A smart walking stick for elderly users built on an ESP32. An MPU6050 motion sensor and an FSR402 grip sensor let it count steps, recognise holding, leaning, standing up and walking, and detect real falls from the 3D tilt angle and how long the stick stays down. When a fall happens it beeps, lights an LED and sends an alert to a Wi-Fi app, and it calibrates to each user's grip automatically.

DMT322

Digital Image Processing Technique

Unity C# YOLO Unity Sentis Computer Vision
Unity simulation of the Na Luang junction with roads, a roundabout, buildings and cars
Computer Vision · Group project (2 members)

Smart Traffic Light Simulation

A digital twin of the Na Luang three-way junction in front of KMUTT, built in Unity, where virtual CCTV cameras feed a YOLO model running in Unity Sentis to count and classify the cars waiting on each road. Instead of a fixed timer, each green light lasts 1.5 seconds per waiting car, between 3 and 15 seconds. Of the four detectors we compared, YOLOv11-Nano counted most accurately (MAE 0.18 at 58 FPS), and at peak hours the adaptive lights cut the average delay from 81.1 to 34.1 seconds per car (58% less) and moved 66% more cars per minute.

MMI320

Medical Instruction Media Technology

Unity Vuforia Augmented Reality C# Android
HemoLens AR in Unity: blood-group card markers with 3D blood-cell models
AR Learning App · Group project (3 members)

HemoLens AR

A marker-based AR app that teaches blood-group compatibility, made for people who want to donate blood but don’t know whose blood theirs matches. Scanning one of eight blood-group cards shows a 3D model of that group’s blood cells, along with which groups it can give blood to and receive from, and moving the phone lets you look at the model from every side. Built in Unity with the Vuforia SDK for Android.

MDT312

Web Programming

HTML CSS JavaScript jQuery Node.js Express MySQL
Home page of the first portfolio website with a photo and the heading I'm Thanyathorn
Personal Website

Portfolio Website

My first personal portfolio site, hand-coded for MDT312 in HTML, CSS and jQuery and published on GitHub Pages. It has Home and Resume, Projects and Contacts pages with a fixed navigation bar, smooth scrolling and active-section highlighting, plus links to my course exercises: two recipe pages, a web board, form validation and a JavaScript square game.

UnNiYam Shop home page with a fashion photo collage, search bar and vouchers
Full-Stack Web App

UnNiYam Shop

An online fashion store built with HTML, CSS and vanilla JavaScript on a Node.js and Express REST API with a MySQL database. Customers browse by category, search, collect vouchers and check out with saved addresses and PromptPay, COD or card; sellers add and edit products with up to five images; and admins manage users and approve seller requests. The demo runs entirely in your browser with sample data, so you can try the customer, seller and admin roles.