Pests Sound Classification IoT Device

A leading Germany based consumer goods MNC wanted to build an R&D prototype device for classifying pests sound to compute the density of pests.

Geography

Germany

Domain

IoT Product Engineering and Cloud Software Engineering

Project Type

IoT, Cloud Platform

Objective

Build Pests Sound Classification device's hardware, embedded software and enclosure ID. Build audio ML model from collection of 1000+ audio reference samples. Build backend Cloud Dashboard application which manages all Pests Sound Classification devices and collects the data via HTTP and MQTT protocols. Build the Cloud Dashboard to show the density of pests vs time vs each device in 3D graph for users.

Approach & Solution

  • Designed the device's embedded hardware based on XENO+ ML SOM module.
  • Developed the embedded software of the devices
  • Built the Tiny ML model for audio sound classification.
  • Developed the backend Cloud Dashboard application on top of our existing Customizable Dashboard app code base.

Business Value

  • Avoided manual density calculations. Accuracy of the classification helped in fine tuning their product roadmaps better
  • Brings reduced time frame for pests density calculation testing.