| The Sterile Insect Technique (SIT) is a biological pest control strategy based on the release of sterilized males of the target insect species. The SIT workflow includes mass rearing, sex sorting, sterilization, and release. The release of females must be avoided, as they are hematophagous and potential vectors of disease transmission, making accurate sex separation a critical step in Aedes SIT programs. Currently, sex sorting is performed using mechanical methods that achieve error rates below 1%; however, the unintended release of female mosquitoes still poses significant risks, requiring additional verification prior to field deployment. In this work, we propose an AI-based verification system to screen SIT release containers before release. The proposed approach exploits the distinct wingbeat acoustic signatures of male and female mosquitoes and extends previous flight-sound methods by integrating acoustic analysis with computer vision for improved sensitivity under realistic container conditions. To ensure consistent and analyzable signals, an innovative stimulation setup based on controlled CO 2 pulses was designed to induce synchronized wing flapping and generate repeatable acoustic activity. Experimental results demonstrate that the proposed multimodal vision–acoustics approach substantially improves female detection performance, achieving an F-Score of 0.99 compared to approximately 0.80 obtained by previous experimental AI methods. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.