Identification of Flying Insects in the Spatial, Spectral, and Time Domains with Focus on Mosquito Imaging

Abstract

Insects constitute a very important part of the global ecosystem and include pollinators, disease vectors, and agricultural pests, all with pivotal influence on society. Monitoring and control of such insects has high priority, and automatic systems are highly desirable. While capture and analysis by biologists constitute the gold standard in insect identification, optical and laser techniques have the potential for high-speed detection and automatic identification based on shape, spectroscopic properties such as reflectance and fluorescence, as well as wing-beat frequency analysis. The present paper discusses these approaches, and in particular presents a novel method for automatic identification of mosquitos based on image analysis, as the insects enter a trap based on a combination of chemical and suction attraction. Details of the analysis procedure are presented, and selectivity is discussed. An accuracy of 93% is achieved by our proposed method from a data set containing 122 insect images (mosquitoes and bees). As a powerful and cost-effective method, we finally propose the combination of imaging and wing-beat frequency analysis in an integrated instrument.

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

“…In recent years, we are witnessing an upsurge of interest in technologically advanced devices as applied to automatic insect detection, counting, and identification [1]. There are mainly three major approaches: (a) optical counters attached to the entrance of traps that target specific pests using lures (pheromones in the case of lepidoptera [2] and palm pests, soil arthropods [3,4] or scents and CO 2 in the case of mosquitoes [5]), (b) camera-based traps that take a picture of their internal space [6][7][8][9][10][11][12][13][14], and (c) near infrared sensors [15] and lidars that emit light covering a volume of space of the open field and registering the backscattered wingbeat signal of flying insects [16][17][18]. All approaches have advantages and disadvantages depending on the application scenario.…”

“…This work adopts the third approach as it is best suited for crawling/walking Arthropoda encountered in urban environments. There is a large corpus of previous approaches on cameras embedded in insect traps (see [6][7][8][9][10][11][12][13][14] and the references therein). Their role is to either report a picture to a server and let a human observer discern the number and the species of the captured insects [14], or proceed into processing the captured image to detect [10,12] and automatically count and/or identify insects [7][8][9][10][11]13,19].…”

“…There is a large corpus of previous approaches on cameras embedded in insect traps (see [6][7][8][9][10][11][12][13][14] and the references therein). Their role is to either report a picture to a server and let a human observer discern the number and the species of the captured insects [14], or proceed into processing the captured image to detect [10,12] and automatically count and/or identify insects [7][8][9][10][11]13,19]. In the former case, the gain is the reduction of cost and manpower required to visit the network of traps and deliver the photos.…”