AI-Powered Mobile Image Acquisition of Vineyard Insect Traps with Automatic Quality and Adequacy Assessment

The increasing alarming impacts of climate change are already apparent in viticulture, with unexpected pest outbreaks as one the most concerning consequences. monitoring pests is currently done by deploying chromotropic and delta traps, which attracts insects present production environment, then allows human operators to identify count them. While these traps still mostly through visual inspection winegrowers, smartphone image acquisition those starting play a key role assessing pests’ evolution, well enabling remote taxonomy specialists better onset outbreaks. This paper presents new methodology that embeds artificial intelligence into mobile devices establish use hand-held capture insect for detection deployed vineyards. Our combines different computer vision approaches improve several aspects quality adequacy, namely: (i) focus validation; (ii) shadows reflections (iii) trap type detection; (iv) segmentation; (v) perspective correction. A total 516 images were collected, divided three datasets manually annotated, order support development validation functionalities. By following this approach, we achieved an accuracy 84% detection, 80% 96% shadows/reflections (for respectively), mean Jaccard index 97% trap’s segmentation.