Edge Computing for Vision-Based, Urban-Insects Traps in the Context of Smart Cities.

Ioannis Saradopoulos,Antonios I Konstantaras,Emmanuel N Antonidakis,Stavros Ntalampiras,Ilyas Potamitis

doi:10.3390/s22052006

Abstract

Our aim is to promote the widespread use of electronic insect traps that report captured pests to a human-controlled agency. This work reports on edge-computing as applied to camera-based insect traps. We present a low-cost device with high power autonomy and an adequate picture quality that reports an internal image of the trap to a server and counts the insects it contains based on quantized and embedded deep-learning models. The paper compares different aspects of performance of three different edge devices, namely ESP32, Raspberry Pi Model 4 (RPi), and Google Coral, running a deep learning framework (TensorFlow Lite). All edge devices were able to process images and report accuracy in counting exceeding 95%, but at different rates and power consumption. Our findings suggest that ESP32 appears to be the best choice in the context of this application according to our policy for low-cost devices.

Highlights

Our aim is to promote the widespread use of electronic insect traps that report captured pests to a human-controlled agency
Smart cities rely on a range of technologies—including artificial intelligence (AI), the internet of things (IoT), and wireless connectivity solutions to provide social services that promote quality of life and sustainability to their citizens
Images of insects found on the internet or in online biodiversity databases are not suitable for training devices operating in the field, as they are of high-quality and close-focus, which does not match the pictures taken from the internal space of operational traps in the field

Summary

Introduction

Our aim is to promote the widespread use of electronic insect traps that report captured pests to a human-controlled agency. We present a low-cost device with high power autonomy and an adequate picture quality that reports an internal image of the trap to a server and counts the insects it contains based on quantized and embedded deep-learning models. Sensor technology and AI practices that process these sensors can leverage detection and density estimation of creatures that have attained the pest status in daily practice. This includes rodents spreading through a network of buildings, stinging insects that carry vector-borne diseases (mosquitoes, biting midges), wood-boring insects that can inflict structural damage to wood (termites, woodboring beetles in urban greenery), sanitary problems in hospitals, schools, metro lines (cockroaches), domestic health threats (bed bugs), or simple annoyance only by the insect presence (ants in houses, clothes moths, spiders, millipedes, centipedes). Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations

Objectives

Methods

Results

Conclusion