Projects

A passive detection system capable of identifying Drones through the characteristic ELF magnetic signatures generated by drone motors.

SolarMamba is a multi-modal deep learning architecture for intra-hour solar irradiance forecasting that adaptively fuses all-sky imagery with meteorological time-series through LadderFusion — a hierarchical, externally-conditioned channel gating mechanism that learns when to trust visual features based on atmospheric context.

A novel semantic segmentation methodology integrating RGB images with optical flow inputs for robust performance under adverse weather conditions, achieving 0.8034 mean IoU in fog, rain, and overcast scenarios.

Agent Based Modeling for Human-Animal Behavior and Interaction. Exploring Elephant and Baboon datasets for movement pattern analysis using machine learning and mathematical perspective.

Novel GAN framework denoises ELF electromagnetic signals from drone motors, outperforming traditional methods and boosting CNN detection accuracy by 24%.

A novel framework integrating cyclostationary signal processing with deep learning for detecting shaft misalignment in rotating machinery, achieving 96% accuracy using SCD-based CNN classification.

An AI-enabled radar detection system utilizing CST-simulated RCS signatures and deep learning to identify body-worn improvised explosive devices with 96% accuracy at standoff distances.