Neural Network Optimization Algorithm for Concrete Crack Image Recognition

Abstract

Concrete structure is one of the most common materials in modern architecture and infrastructure construction. However, cracks will inevitably appear in the use of concrete structures, which is an important factor affecting the safety and service life of structures. Traditional crack analysis relies on manual participation, which reduces work efficiency and also has human influence. Based on the development of artificial intelligence, image recognition with neural network has become a desirable solution. In order to solve the problem of concrete crack image recognition, an optimization scheme using neural network is proposed. By introducing advanced neural network model and combining with image processing technology, the algorithm realizes the automatic recognition and classification of concrete cracks. In the process of algorithm design, the structure of neural network is optimized, including the network structure, overfitting and the selection of additional modules, so as to improve the recognition and classification accuracy. In addition, aiming at the imbalance of data sets in practical applications, data enhancement technology is introduced to effectively improve the generalization ability of the model. The experimental results show that the proposed neural network optimization algorithm has achieved remarkable performance improvement in the task of concrete crack image recognition.