Mosquito-borne diseases pose a major public health challenge and require effective, scalable surveillance to guide targeted interventions. Existing monitoring techniques, ranging from manual morphological identification to acoustic, optical, and spectroscopic sensing, remain constrained by environmental sensitivity, labor demands, and limited ground-truth validation. Here, we present a fully autonomous, field-deployable platform, called automated intelligent mosquito sentinel (AIMS), integrating distributed mosquito monitoring outposts (MMOs) and a centralized analysis center (AC) for scalable, non-invasive mosquito surveillance. AIMS employs an adaptive event-triggering mechanism, optimized through feature engineering of colour and texture pairs, to enable energy-efficient detection with zero missed events and a false-positive rate below 1%. At the analytical level, a hierarchical gated residual network performs multitask classification of taxonomy and sex with accuracies of 99.51% at species and 98.02% for sex, demonstrating interpretable, biologically meaningful attention patterns. The self-powered architecture, robust wireless data transmission, and large-scale field dataset underpin reliable operation across diverse ecological settings. Together, these results show that AIMS can support scalable and sustainable mosquito surveillance and may also be useful for broader entomological monitoring and public health applications. Nuofei Lin and colleagues report an autonomous intelligent mosquito surveillance system designed for field deployment. The system integrates event-triggered standardized imaging with an interpretable hierarchical classification framework to enable real-time, non-invasive mosquito monitoring
Lin et al. (Thu,) studied this question.