A Pixel-wise Segmentation Model to Identify Bur Chervil (Anthriscus caucalis M. Bieb.) Within Images from a Cereal Cropping Field

Abstract Because of insufficient effectiveness after herbicide application in autumn, bur chervil ( Anthriscus caucalis M. Bieb.) is often present cereal fields spring. A second reason for spreading the warm winter Europe due to climate change. This weed continues germinate from autumn To prevent further spreading, a site-specific control spring reasonable. Color imagery would offer cheap and complete monitoring entire fields. In this study, an end-to-end fully convolutional network approach presented detect within color images. The dataset consisted images taken at three sampling dates 2018 wheat one date 2019 rye same field. Pixels representing were manually annotated all After random image augmentation was done, Unet-based neural model trained using 560 (80%) sub-images (training images). power different evaluated 141 (20%) (100%) (test Comparing estimated plants test Intersection over Union (Jaccard index) showed mean values range 0.9628 0.9909 2018, value 0.9292 2019. Dice coefficients yielded 0.9801 0.9954 0.9605