Personal Guardian Safety System (PGSS)

A wearable fall detection device built on ESP32, utilizing an on-device Convolutional Neural Network (CNN) for real-time fall detection. The system features Telegram emergency alerts with GPS location tracking and a dedicated web dashboard to log fall history.

🚀 Key Features

• Real-Time Edge AI: Executes a trained CNN model directly on the ESP32 for immediate fall detection without relying on continuous cloud processing.
• Telegram Emergency Alerts: Automatically dispatches instant notifications containing real-time GPS coordinates to designated contacts when a fall is detected.
• Gmail Emergency Alerts: Automatically dispatches instant notifications containing real-time GPS coordinates to designated contacts when a fall is detected.
• Web Dashboard Integration: Logs and displays fall history on a centralized web interface for continuous monitoring and review.
• Custom Memory Allocation: Utilizes a custom partition scheme to efficiently accommodate the embedded machine learning model.

🛠️ Technologies & Hardware

• Microcontroller: ESP32
• Development Environment: PlatformIO
• Languages: C, C++
• Notifications: Telegram Bot API and gmail
• Machine Learning: Embedded CNN

📂 Repository Structure

• main.cpp — Core application logic, sensor data acquisition, and inference execution.
• pgss_model.h — The exported, on-device CNN model header file.
• partitions.csv — Custom memory partition table defining space for the ESP32 application and ML model.
• platformio.ini — PlatformIO project configuration, build environments, and library dependencies.
• LICENSE — MIT License documentation.