Аннотация
Seasonal monitoring of cultivated land for weeds with automatic classification of crops and weeds provides farmers with a significant return on investment, as it saves fertilizer, herbicides and time. This paper discusses the problem of semantic image segmentation using Convolutional Neural Network (CNN) architectures. In the implementation of the CNN architectures RGB format images are used. A comparative analysis of the application of the proposed approaches and their estimated performance indicators obtained as a result of experimental studies is carried out.