Neural Networks & Cognitive Computing

An IoT and cloud-based telerehabilitation framework using RGB-Depth sensors for remote patient monitoring and rehabilitation assessment. The system captures 3D skeletal data and RGB video to analyze exercise quality, range of motion, and treatment adherence, providing real-time feedback to patients and progress reports to therapists via secure cloud infrastructure for at-home rehabilitation programs.

A multimodal IoT-enabled framework for human action recognition in UAV/drone-based crowd and multi-person scenarios. The system integrates aerial RGB video with IoT sensor data through deep learning pipelines to recognize human actions from elevated viewpoints, addressing challenges of scale variation, occlusion, and dynamic camera motion in real-time surveillance and search-and-rescue operations.

A multi-modal feature fusion pipeline for myocardial infarction detection from echocardiogram and MRI sequences using deep learning. The system extracts complementary features from both imaging modalities, fuses them through attention-guided mechanisms, and classifies cardiac abnormalities with high sensitivity and specificity, enabling early detection of heart attacks and improved patient outcomes in clinical cardiology.