Computer Vision-Based Ovitrap for Dengue Control

Abstract

According to information from health institutions, the deadliest animal in the world is the Aedes aegypti mosquito. Regarding Mexico, in 2020, 9,000 infections and 580 deaths caused by dengue were registered. Veracruz, Tabasco, Guerrero, Nayarit and Tamaulipas are the states that represent 61% of the infections. Currently, one of the ways to combat it is to identify the areas with the highest risk of contagion by installing and reviewing ovitraps. The review and counting process is carried out manually by specialised personnel, every week. This leads to generating a certain degree of uncertainty in the information collected, in addition to the excessive consumption of resources. In some countries, such as Malaysia, Indonesia, Brazil, among others, attempts have already been made to automate the process of collecting and counting eggs. Some have used embedded systems, while others have focused on the implementation of techniques derived from artificial intelligence. However, the results presented are around controlled environments (laboratories). In this work, an ovitrap prototype is presented that uses Raspberry Pi technology, integrated with software based on artificial vision techniques, which allows the images obtained from inside the ovitrap to be analyzed, this by means of segmentation and a simple counting of the eggs deposited by the Aedes aegypti mosquito. At the moment, preliminary results are satisfactory, since they are based on more than a hundred images in real environmental conditions.