Skip to content
Deployment in the field (light turns on upon detection)

Insect Detect night

Insect Detect night is an open-source system used for non-invasive automated monitoring of flower-visiting nocturnal insects in their natural habitat. It integrates a monochrome camera with adaptive infrared illumination and a grayscale deep-learning object detection model that runs directly on the device in real time. When an insect is detected, the system triggers an LED ring and an RGB camera to record images suitable for species identification in post-processing. This event-based workflow enables continuous, non-lethal, and low-disturbance monitoring of nocturnal flower visitors.

Deutsch

Example image taken by the device
On-device pipeline
  • Host: Raspberry Pi Zero 2 W
  • Camera: Luxonis OAK-D Pro (12MP RGB sensor, 1MP monochrome sensors; with integrated IR light)
  • Power manager: Witty Pi 4 L3V7
  • Batteries: 88Wh Li-Ion
  • Power: USB Battery Pack
  • Power consumption: ~4.2 W; can run the whole night when powered with battery
  • Extra components: light sensor to measure ambient light
  • No attractant
  • On-device processing: detection, tracking
  • Components price: ~830 €

Contact:
Jasmin Krebs
jassi.krebs@gmx.de
Developed at: German Centre for Integrative Biodiversity Research Halle-Jena-Leipzig (iDiv)

github.com/JasminKrebs/insect-detect-night