A Novel Method for Studying Mosquito Oviposition Behaviour Using Computer Vision and Deep Learning Algorithm

Understanding the oviposition behaviour of mosquitoes is crucial for developing a vector surveillance program and a control strategy. To study this behaviour, in situ observation is one of the ways to determine the details of oviposition preference. However, this method of data collection is time-consuming and labour-intensive, and the presence of human observers often causes odour nuisance, which can lead to bias. We demonstrated a novel method that able to study this behaviour, which we named Automatic Mosquito Oviposition Study System (AMOSS) that automatically detects and measures mosquito oviposition activity and collects data without human intervention. The system consists of a microcomputer with an infrared camera that records time-lapse video in a dark environment, and a post-record processing component for detecting the activity by using a deep learning algorithm. We used the system to study the oviposition activity of Aedes mosquitoes on a disposable mask and the result was consistent with the standard oviposition testing - egg counting bioassay. This technology could be an additional tool to determine mosquito preference for a particular substrate, which is very helpful in developing a push-and-pull strategy for mosquito control.