Skip to main content
AMCA26 Main banner
Final Program


Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.

Artificial Intelligence Applications

Artificial Intelligence Applications

Artificial Intelligence Applications

276 - Deployed Accuracy of IDX: A CNN based image recognition system

Thursday, March 26, 2026
2:05 PM - 2:15 PM PDT
Location: B116
Abstract: Controlling vector-borne diseases relies on accurate mosquito species identification, which is often challenging due to morphological similarities. IDX is a system capable of identifying the species of field-caught adult mosquitoes and ticks using convolutional neural networks (CNNs). IDX has been used by over 60 organizations since its market introduction in 2021. Since that time, users have imaged over 267,000 mosquitoes and provided a species identification for 57,000 specimens (nearly 21%). The most recent algorithm released in May 2025 supports 71 mosquito species and 32 tick species with a development accuracy of 95% and 96% respectively. Although multiple groups have demonstrated high development accuracy using computer vision techniques for mosquito ID, deployed accuracy on new specimens has been less compelling. In this study we analyzed deployed accuracy of the most recent algorithm used in IDX by comparing algorithm ID to the ID provided by our users on all available field-caught specimens that had not been used for algorithm development. In total there were 29,861 specimens accounting for 68 of the 71 mosquito species included in the algorithm. 75% of specimens were from the United States, with the remaining 25% primarily from Africa, Europe and Oceania. The overall deployed accuracy was 79%. At the genus level, recall was 78%, 82% and 83% for Aedes, Culex and Anopheles respectively. This study demonstrates the feasibility of translating mosquito image recognition based technologies to practical field tools. Translation of these algorithms to other product concepts such as a rapid mosquito identification system and a smart trap is currently in progress.