Pulindu Vidmal's Projects
AgroMate: Revolutionizing farming with IoT-driven precision. Optimize water usage, monitor soil health, and protect crops from wildlife threats. Transform agriculture sustainably.
Automate and enhance your clothesline experience with this Arduino-based system. It automates the retraction process triggered by rain detection, protecting clothes during inclement weather. The system offers remote data monitoring via Thingspeak and convenient control through a mobile app.
AutoGluon simplifies ML model development by automating tasks like model selection, tuning, and deployment. With support for various data types and seamless AWS integration, it enables easy prototyping and scalable deployment. This repository contains the core functionality and resources for utilizing AutoGluon effectively.
[ICRA'23] BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation
This repo includes a collection of my experiments, assignments related to of EN3160 Image Processing and Machine Vision module.
facial sketching tool for enhanced police and private investigations using advanced GANs and ML algorithms
This project develops a deep learning model using CNNs and Residual Blocks to predict facial key-points. The application spans from emotion recognition in AI to driver monitoring systems.
The Five-Band Audio Equalizer project, part of the Laboratory Practice module in semester 3, aims to develop and implement a graphic equalization circuit.
This repo presents an automated H-bridge assembly system using a metal ring and 6 transistors, designed to enhance efficiency and quality in motor control applications.
Detect holes in a video feed and calculates the angle of rotation of the ring needed to align one of the holes to the initial point
This project explores the application of auto-encoders for image noise reduction. Auto-encoders, a type of neural network, excel at learning to compress and reconstruct data, making them effective for tasks such as denoising images.
This repository is dedicated to object localization model training with the EfficientNet architecture. It employs the Albumentations library for data augmentation and the Timm library for accessing pre-trained EfficientNet models.
This repository contains my solutions to the problems on Project Euler.
This repository implements a U-Net based approach for semantic segmentation of aerial images, using the Aerial Semantic Segmentation Drone Dataset. The model is designed to classify each pixel in the images into one of 23 categories, improving the understanding of urban scenes for safer autonomous drone navigation.
The application extracts significant frames from a video, generates descriptions for these frames, and then creates a summary of the video content.
This collection is dedicated to exploring and understanding key concepts in signal and systems theory. It includes experiments and assignments related to the courses EN1020: Circuits, Signals, and Systems, EN2063: Signals and Systems, and EN3551: Digital Signal Processing.
Computer Vision system for identifying 3D-objects and identifying colors using Deep Learning and OpenCV. Made for SLRC 2024.
The Medibox project, initially developed for the EN2853 - Embedded Systems & Applications module, is a smart device that reminds users to take medication on time and monitors temperature and humidity for proper storage conditions.
Sparkle Robot, a versatile robotic platform powered by Arduino to fulfill the EN2533: Robot Design and Competition tasks
This project focuses on training a Deep Convolutional Neural Network (CNN) for the task of traffic sign classification. With the goal of enhancing the safety of autonomous vehicles, the CNN is designed to accurately identify and classify different types of traffic signs commonly found on roads.
This repo documents a project undertaken as part of the EN2111 Electronic Circuit Design module at the University of Moratuwa, focusing on the implementation of a UART communication link between two FPGA boards.