COMPARATIVE ANALYSIS OF MODERN NEURAL NETWORK ARCHITECTURES FOR COMPUTER VISION TASKS

Nazipov Rustam Salavatovich

This article presents a comparative analysis of three modern neural network architectures for computer vision tasks: Convolutional Neural Networks (CNNs), ResNet, and YOLO. The key features, advantages, and limitations of each architecture are examined. The practical significance and future development prospects of neural networks in the field of computer vision are discussed, including the development of hybrid models, the use of transfer learning methods, and integration with classical approaches. The importance of further research to improve the efficiency, adaptability, and interpretability of neural networks in solving a wide range of computer vision problems is emphasized.

Keywords:computer vision, neural networks, convolutional neural networks, ResNet, YOLO, neural network architectures, deep learning, interpretability, transfer learning, hybrid models