TOWARDS AN EFFECTIVE DRONE-BASED TECHNOLOGY TO COMBAT MOSQUITO BREEDING SITES

Resumo

Drones have become an important technological tool to support health surveillance teams in locating
and eliminating mosquito breeding sites in urban areas, since they allow the acquisition of aerial images with high
spatial and temporal resolutions. However, such images are often analyzed through manual processes, consuming
a lot of time in the inspections. Thus, researchers from different parts of the world have proposed computer vision
technologies to enable automatic identification of mosquito breeding sites with the use of drones. However, despite
the advances achieved, many of these technologies have some limitations that difficult their use in practical
situations. This study is focused on such technologies aiming to investigate their potential, limitations and
scalability. The literature review carried out shows that some proposed computer vision approaches could easily
compose low-cost softwares to be used in tasks aimed at combating mosquito breeding sites with the use of drones,
optimizing time and human resources. Despite that, works addressing software development from the proposed
computational approaches are very rare. Other limitations found in the present study are the non-identification of
water in the detected targets (suspicious objects and scenarios) and the non-provision of accurate geolocation of
them. In addition, in the case of Brazil, the lack of public policies to adequately support the use of the investigated
technologies can also be seen as a limitation.