Abstract
The following quote was obtained by the news editors from the background informa tion supplied bythe inventors: “Identification of plant traits in digital image ry has numerous applications. One exampleapplication is detection, classificati on, and/or segmentation of unwanted and/or invasive plants, such asweeds. Anoth er example application is detection, classification, and/or segmentation of plan t disease. Yetanother example application is detection, classification, and/or segmentation of other types of plant traits,such as plant genus or species (eit her of which may fall under the definition of plant “type”), phenotypictraits ( e.g., gender, strain), and so forth. Various types of machine learning models ca n be trained tosegment and/or recognize various types of plant traits in digita l images. Convolutional neural networksare one popular example. However, the ac curacy of these machine learning models depends largely on theamount of trainin g data used to train them. Manually annotating training images on a pixel-wise b asisand/or using bounding shapes can be prohibitively costly. Moreover, images of plants having some traitsmay not be as widely available or easily acquired a s images of plants having other traits.”
NETL
NSTL