Automated lab identification and sorting (Rapid-IDX) for mosquito surveillance

点击翻译按钮获取中文摘要

Overview. The proposed work will advance state of the art mosquito surveillance technology through development of an automated species identification and sorting system, Rapid-IDX. The proposal builds on Phase I results, during which the mechanical subcomponents for isolation, separation, and transportation, multi-angle optics system, and prototype algorithm were designed and validated. For phase II, the optics and mechanical systems designed will be integrated into functional prototypes for a minimum viable product (MVP), an image database will be developed to focus algorithm development on key medically-relevant mosquito species, and iterative functionality and usability testing will be conducted with potential customers. Intellectual Merit. Rapid-IDX will allow for rapid data generation of high-resolution, multi-angle images of mosquito specimens for training of Convolutional Neural Networks (CNNs) for identification; the large and robust dataset enabled with this technology will help push Vectech’s technology towards more nuanced and powerful algorithms. This phase II work centered on StarAttGAN represents the state of the art with regard to generative models for dataset augmentation. These developments will enable addressing a critical issue in computer vision: that of sampling bias in vast noisy, fine-grained, openset multiclass classification problems. Broader Impacts. The proposed work supports United States critical public health infrastructure and public health literacy in the face of growing incidence of vector borne disease as a result of climate change, globalization, and other factors. The system will provide Vector control organizations (VCOs), which protect local communities in the US from disease-carrying insects, with more robust capacity in vector surveillance data gathering, a critical component of integrated vector management. Additionally, US Armed Forces face a significant threat of vector-borne diseases around the world, making the Department of Defense one of the largest customers for vector surveillance and control products. We are seeing emerging competitors in the pest information space in Germany, Israel, China, and South Korea (see commercialization plan). Investment in Vectech ensures there’s an American company to compete in the global market. We will collaborate with leading US-based MCOs for development and optimization of this product, ensuring product-market fit. Furthermore, the technical team leading this phase II proposal represents several underrepresented groups in STEM, including women, BIPOC, and the LGBTQ community.