AI / ML2023completed

Sinhala Character Predictor

An advanced machine learning project focused on optical character recognition (OCR) for the Sinhala script. This project implements neural network models to recognize and predict Sinhala characters from image inputs, addressing the unique challenges of non-Latin character recognition and script-specific pattern matching.

PythonMachine LearningTensorFlowOpenCVJupyter
View on GitHub

2023

Year

6

Features

6

Technologies

Problem
  • ·Limited availability of labeled Sinhala character datasets
  • ·Complexity of Sinhala script with diacritical marks
  • ·Balancing model accuracy with inference speed
  • ·Handling similar-looking characters effectively
Solution
  • ·Created custom dataset through data collection and annotation
  • ·Implemented specialized preprocessing for Sinhala script
  • ·Used transfer learning to improve model performance
  • ·Applied data augmentation to build robust models

Key Features

Convolutional neural network for image classification

Image preprocessing pipeline with normalization

Data augmentation for improved model robustness

Multi-layer neural network architecture

Real-time character prediction capability

Model evaluation with precision, recall, and F1 metrics

Technologies

PythonTensorFlowKerasOpenCVNumPyScikit-learn

Learnings

  • Developed proficiency in neural network architecture design

  • Gained expertise in image preprocessing and augmentation

  • Learned language-specific ML challenges and solutions

  • Mastered model evaluation and performance metrics

Highlights

Neural Network Model

OCR Implementation

Character Recognition

Next Project

MicroBankingSystem Backend

Robust backend system for banking operations with database architecture. Group project involving microservices and database design.