This thesis focuses on the development of a software tool for detecting and recognizing road signs using a convolutional neural network (CNN). The work involves an analytical review of existing systems and algorithms for detecting and recognizing road signs in real scenes, highlighting their advantages and disadvantages. An overview of the technical parameters of ready-made sign recognition systems is also provided. To solve the problem, the combination of two algorithms is proposed: shape detection using the Gabor filter and the MSER (Maximal Stable Extreme Regions) method. The process of recognizing road signs using artificial and CNNs is described. The use of training samples GTSRB and GTSDB is discussed, and a software implementation based on the Keras and OpenCV libraries is presented, achieving a high percentage of road sign identification. The thesis concludes with a discussion of the practical value of the results, which can be applied to recognizing road signs and other objects in video surveillance systems and other areas related to ensuring road safety. The work was defended at the Scientific and Technical Conference "Information Models, Systems and Technologies" at Ternopil Ivan Pul'uj National Technical University (December 7-8, 2022).This thesis focuses on the development of a software tool for detecting and recognizing road signs using a convolutional neural network (CNN). The work involves an analytical review of existing systems and algorithms for detecting and recognizing road signs in real scenes, highlighting their advantages and disadvantages. An overview of the technical parameters of ready-made sign recognition systems is also provided. To solve the problem, the combination of two algorithms is proposed: shape detection using the Gabor filter and the MSER (Maximal Stable Extreme Regions) method. The process of recognizing road signs using artificial and CNNs is described. The use of training samples GTSRB and GTSDB is discussed, and a software implementation based on the Keras and OpenCV libraries is presented, achieving a high percentage of road sign identification. The thesis concludes with a discussion of the practical value of the results, which can be applied to recognizing road signs and other objects in video surveillance systems and other areas related to ensuring road safety. The work was defended at the Scientific and Technical Conference "Information Models, Systems and Technologies" at Ternopil Ivan Pul'uj National Technical University (December 7-8, 2022).