Real-Time Rodent Detection & Monitoring System

The Challenge

Processing real-time video on a microcontroller while simultaneously fusing three sensor modalities (PIR, ultrasonic, mmWave) and rendering a responsive graphical UI required careful core partitioning. The Cortex-M7 core handles video capture and display rendering via the DMA2D+LTDC pipeline, while the Cortex-M4 core manages sensor polling and decision logic — preventing any single workload from starving the others. Achieving ~24 FPS without frame tearing on the LCD required a double-buffered DMA2D transfer strategy with precise interrupt synchronization.

Technical Overview

A dual-core STM32H747I-DISCO based embedded system designed for real-time rodent detection and monitoring. The project integrates an onboard camera for live video capture and LCD streaming at ~24 FPS using DMA2D double buffering. Multiple sensors (PIR, ultrasonic, mmWave) are fused for accurate rodent activity detection. The system features a custom LVGL-based UI for local monitoring and Ethernet connectivity for remote data visualization and control.

Key Features

• Real-time video streaming at ~24 FPS using DMA2D double buffering
• PIR, ultrasonic, and mmWave sensor fusion for reliable rodent detection
• LVGL-based graphical UI for local on-device monitoring
• Ethernet connectivity for remote data logging and live viewing
• Dual-core task partitioning: Cortex-M7 for vision, Cortex-M4 for control
• Low-latency DMA2D + LTDC display pipeline with no frame tearing