Image translation to various art styles using DiffusionCLIP
The aim of Neural Style Transfer is to give the Deep Learning model the ability to differentiate
between the style representations and content image. Our take on generating new variations of existing art pieces
Accessibility focused web-component for assisting dyslexic readers
A web-component and an e-reader utlising the web-component to modify
html elements associated with any webpage using an interactive toolbar
Event based monitoring of parameter change for ROS2-IGTL bridge
Direct communication between ROS and open-IGTL. Updated parameters sent to IGTL
when ROS2 parameter values are updated or new parameters are created.
Heap based snapshot creation for millisecond high frequency trade data
Efficient generation of limit order book from millisecond trade data using a custom heap based algorithm.
SLAM based on Bayesian estimates
TSimultaneous localisation and mapping of rover in an enclosed environment based on laser sensor readings - using bayesian estimates.
Pneumonia Detection in X-Rays | Runner-up Astrazeneca AI Challenge
Tackled as an object detection problem, with only one type of image to detect. Yolo-v3 from Darknet was used.
Achieved an IoU of 46.97% and MAP value of 35.44%.
Non-Contact Vibration Analysis through video | Machine Design Section, IIT Madras
Investigated video cameras as an inexpensive way for remote vibration analysis; built in Python using OpenCV.
Utilized scale-invariant features and object tracking algorithms to extract the motion signal from video frames
Indian Summer Rainfall Prediction | Heat Transfer Lab, IIT Madras
Developed an ML model for prediction of Indian rainfall on monthly and seasonal time scales.
Used RNNs to forecast rainfall for June to Sept (ISMR) by feeding data on rainfall and sea surface temps
Livestock Management Using Unique IDs | Runner-up AgHack 2020
Ideated, pitched, & showcased Dairy Folk: A start-up focusing on improving livestock management using unique IDs.
Enabled continuous health monitoring of livestock by decentralizing storage of data for enhanced transparency
Quality Prediction in Iron Ore Mining, IIT Madras
Employed random forest classifiers, neural networks and XGBoost in an ensemble to predict the
output Silica concentrate based on input ore properties, ore pulp condition, air flow rate & other process parameters
Moodle Crawler for timely offline back-up creation
Created a web crawler using selenium & bs4 for maintaining offline copy of course documents from Moodle.
Wrote a daemon to refresh the offline version daily; File system updated to match changes in course structure
