Interpreting Lung Ultrasound Scans
Worked on algorithms to interpret various artifacts in lung ultrasound scans that can expedite the diagnosis of COVID-19.
Studied the problem of predicting future head trajectories based on the coorrelation between eye and head movements in a virtual environment.
Detection of objects with reflective properties
Analyzed the effect of depth to detect reflective objects such as mirrors in images.
Scene Text Detection And Recognition
Experimented with a two stage network to detect text from natural images.
Brand Logo Detection
Trained FasterRCNN for detecting logos in natural images. The project also explores methods to detect logos using visual features from a few number of reference images
Representation Learning and Random Projections for Sparse Data
Experimented with various hashing techniques for dimensionality reduction. We also explored multi-task setting for Named Entity Disambiguation (NED).
Classifying Thermographic features for Breast Cancer Screening
Designed an algorithm to automatically classify thermographic images for the presence of breast cancer. We also enable automatic detection of the suspected malignant regions in images taken from various angles.
Distributed Matrix Operations using Spark
Designed an algorithm to compute the inverse of a high dimension symmetric positive definite matrix. We implemented the algorithm in Apache Spark in a distributed environment using HDFS for data storage