AI Research Project

Neural Network-Based System for Early Detection & Intervention of Dysgraphia

An AI-driven platform for early screening and intervention of dysgraphia in children, combining custom neural networks, localized handwriting datasets, and immersive educational tools to support parents, teachers, and therapists.

Domain: Artificial Intelligence & Education
Duration: July 2024 – October 2025
Type: Final Year Research Project

Project Overview

AI-powered screening and intervention for handwriting-based learning disabilities

🎯

The Challenge

Dysgraphia is underdiagnosed in Sri Lanka due to limited awareness, lack of specialists, and absence of localized datasets, causing delayed educational support for affected children.

💡

The Solution

Developed a neural network-based system using a locally collected numeric handwriting dataset and publicly available letter datasets to detect dysgraphia early and accurately.

🚀

The Impact

Enables early screening, gamified intervention, and bilingual AI support for children, parents, teachers, and therapists through a unified platform.

Key Features

Core capabilities of the dysgraphia support platform

🧠

AI-Based Dysgraphia Detection

Advanced CNN models analyze handwriting patterns to identify dysgraphia with high accuracy.

🔢

Numeric & Letter Analysis

Supports numeric dysgraphia using local datasets and letter dysgraphia using curated online datasets.

🎮

Gamified Intervention

Interactive learning activities designed to enhance fine motor skills and handwriting confidence.

🤖

AI Avatar Chatbot

Real-time Sinhala and English chatbot with speech synthesis and lip-sync for child-friendly engagement.

👥

Community Support

Secure discussion spaces for parents, teachers, and specialists to share experiences and guidance.

🔐

Ethical & Privacy Focused

Child-safe design with consent-based data collection and non-diagnostic framing.

Technology Stack

AI, web, and immersive technologies used

💻 Development

Python TensorFlow PyTorch Flask React Tailwind CSS

🗄️ Data & Storage

MySQL Firebase OpenCV Transformers

Project Impact

Key outcomes and achievements

97%Numeric Accuracy
99%Letter Accuracy
8,000+Samples
20+Districts

Interested in AI for Education?

Let’s build inclusive, intelligent systems for early learning support.

Contact Me View More Projects