Recognition of acoustic vortex fields based on a convolutional attention neural network

In this work, we propose a deep convolutional attention neural network to realize the recognition of acoustic vortex fields. Convolutional attention mechanisms are introduced into the network together with the residual idea. The deep neural network is trained and then evaluated, and the performance is compared with those of several typical convolutional neural networks, including AlexNet, GoogLeNet, and ResNet. The results show that the improved neural network model based on the convolutional attention mechanism proposed here has better classification performance and stronger stability. The classification accuracy of the improved model on the whole test set reaches more than 95%, indicating that the model has a stable classification ability. Our work helps further study the detection and recognition of acoustic vortex fields, which will find many applications in scientific research and industry.