COMBINING DRONE-BASED ULTRA-HIGH-RESOLUTION EARTH OBSERVATION DATA WITH AI FOR MOSQUITO LARVAL HABITAT IDENTIFICATION: A SCALABLE METHOD IN MALARIA VECTOR CONTROL

In rural areas, it is difficult to perform larval source management for malaria vector control due to difficulties in identifying target areas. High resolution earth observation data captured by drones were used in Malawi to detect larval habitats. The images were analyzed by a classification model to automate land cover identification. Results show that the model successfully identified larval habitat characteristics with a median accuracy of 98%. Nevertheless, this can only identify potential larva habitats and still requires confirmation on larvae presence through ground sampling. Using this technology could save time identifying larval habitats and better focus malaria vector control efforts.