Traditional heart risk assessment requires clinical equipment and physical contact, making it inaccessible for continuous monitoring. This project addresses the need for non-invasive, real-time cardiac health assessment using just a standard camera.

Leveraged remote photoplethysmography (rPPG) to extract vital signs from facial video. Trained an LSTM-based deep learning model on customized ECG data to enable heart risk prediction from PPG signals extracted via pyVHR and OpenCV.

Built the signal extraction pipeline using pyVHR and OpenCV for real-time facial video processing. Developed a TensorFlow-based LSTM model trained on customized ECG datasets. Created a React dashboard integrated with Firebase for storing, monitoring, and visualizing heart health metrics in real time.

Camera feed → pyVHR signal extraction → OpenCV preprocessing → TensorFlow LSTM prediction → Firebase storage → React dashboard visualization. The system processes video frames in real-time and provides continuous risk assessment updates.