The article presents particulars of developing a plant disease detection system based on analysis of photo-graphic images by deep convolutional neural networks. A original lightweight neural network architecture is used (only 13 480 trained parameters) that is tens and hundreds of times more compact than typical solutions. Real-life field data is used for training and testing, with photographs taken in adverse conditions: variation in hardware quality, angles, lighting conditions, scales (from macro shots of individual fragments of leaf and stem to several rose bushes in one picture), and complex disorienting backgrounds. An adaptive decision-making rule is used, based on the Bayes’ theorem and Wald’s sequential probability ratio test, in order to improve reliability of the results. A following example is provided: detection of disease on leaves and stems of rose from images taken in the visible spectrum. The authors were able attain the quality of 90.6% on real-life data (F 1 score, one input image, test dataset).