Real-time Traffic sign detection for Advanced Driver Assistance Systems

Advanced Driver Assistance Systems (ADAS) aid in driving and can include assistive and safety systems, as well as self-driving agents. However, these systems can be expensive and require significant computational power. Traffic sign detection is crucial to ADAS, but it's challenging due to limited localized data and object detection difficulty.

This research contributes to traffic sign detection and driver assistance by presenting a dataset of 4,000 images with 35 traffic sign types in Sri Lanka using Google Street View images. It addresses challenges such as object detection in varying weather and light conditions and class imbalance problems through data augmentation and transfer learning. Additionally, it presents a smartphone-based driver assistance implementation for real-time traffic sign detection.

Results show that models trained on the collected dataset can detect Sri Lankan traffic signs and converge faster and more accurately using transfer learning strategies. The proposed Android application demonstrates the possibility of creating a real-time traffic sign detection application using smartphones' computational power.

• GitHub (Mobile app) - pasannissanka/DriverAssistance
• GitHub (Object detection) - pasannissanka/driverassistance-cv-core
• GitHub (Data Preparation) - pasannissanka/sign_recognition

Demo

Personal Assistance Minion Robotics

Created an minion based robot for personal assistance using MyCroft AI Open source personal assistance system.

Mediserv - Pharmacy Management System

Service Oriented Computing - Group Assignment

Pharmacy management system was created for CSC 313 - Service Oriented Computing Group project,

Team members were, Premesh Heshan, and Dasun Malshan

This is a cloud-based Pharmacy management system, where pharmacies can register on the platform and customers can create orders by uploading prescription images. Individual pharmacies then can process those prescriptions and then create and dispatch orders through the Admin app.

The system is composed of,

• Spring Boot REST API
  • Spring boot, Spring Security, MongoDB
• ReactJS Admin App
  • ReactJS, Typescript, Tailwind CSS
  • Login credentials
    • Email testuser@mediserv.com,
    • Password 123456
• ReactJS Store App
  • ReactJS, Typescript, Tailwind CSS
  • Login credentials
    • Email testcustomer@mediserv.com,
    • Password 123456

A Hackathon project for IHack 5.0, 24 hour coding Hackathon organized by the ISACA Student Group of UCSC. Participating in Smart cities initiative, our group proposed TransitX, which is a system that is designed to be used by transportation service workers and riders of public transportation.

The system was composed of,

• Android/ iOS app for transportation workers. (TransitX Worker)
• Android/ iOS app for Commuters. (TransitX)
• API.
• Website.