Overview

Elderly Haki (WinForms) is the desktop interface of a comprehensive elderly safety system designed for real-time activity monitoring. This application uses CCTV feeds, AI-based temporal activity recognition, and Firebase syncing to track and detect activities such as standing, sitting, walking, falling, etc.

The system’s intelligence is powered by a 3D Convolutional Neural Network (3D CNN), trained from scratch to analyze short sequences of video frames and recognize motion patterns. Combined with DeepSORT for person tracking, it ensures consistent activity monitoring per individual across time.

Highlights

🧠 Custom-Built AI Model: Trained a 3D CNN from scratch on a custom activity dataset to classify motions over time with 97.03% accuracy.
🎥 CCTV-Based Monitoring: Captures and processes every 5th frame from CCTV feeds, reducing latency and system load.
🔁 Realtime AI Inference: Sends image sequences to a Python WebSocket (ZMQ Dealer) for activity prediction.
👤 Identity-Aware Tracking: Uses DeepSORT to track multiple people across frames, preserving activity history per person.
🖥️ Desktop UI for Caregivers: Built in WinForms, providing a clear view of all active camera feeds with overlayed activity labels.
☁️ Firebase Syncing: Pushes predictions to Firebase Realtime Database, enabling mobile access and alerts.

Tech Stack

C# · WinForms · Python · ZMQ · OpenCV · TensorFlow · Firebase Realtime DB · 3D CNN · DeepSORT · Machine Learning

Design Credits

Angela Marie Avenido

UI/UX Designer, Graphic Artist, Front-End Design

📧 angelamarieavenido113@gmail.com