Intelligent Monitoring System to Assess Plant Development State Based on Computer Vision in Viticulture

Plant health plays an important role in influencing agricultural yields and poor plant can lead to significant economic losses. Grapes are widely cultivated plant, especially the southern regions of Russia. subject a number diseases that require timely diagnosis treatment. Incorrect identification large crop A neural network deep learning dataset 4845 grape disease images was created. Eight categories common typical Black Sea region were studied: Mildew, Oidium, Anthracnose, Esca, Gray rot, White bacterial cancer grapes. In addition, set healthy plants included. this paper, new selective search algorithm for monitoring state development based on computer vision viticulture, YOLOv5, considered. The most difficult part object detection is localization. As result, fast accurate status realized. test results showed accuracy 97.5%, with model size 14.85 MB. An analysis existing publications patents found using “Computer viticulture” technology original promising. developed software package implements best approaches control system viticulture technologies. mobile application practical use by farmer. hardware complex be installed any vehicle. Such will allow real-time vineyards display it map. novelty study lies integration hardware. Decision support adapted solve other similar problems. product commercialization plan focused automation robotization agriculture, form basis adding next software.