Generalized approach of convolutional and pooling layer in image processing using wavelet CNN

Abstract

Spatial and spectral approaches area unit two major approaches for image processing tasks like and beholding. Among several such algorithms, convolutional neural networks (CNNs) have recently achieved significant performance improvement in several difficult tasks. CNNs enable the nation to utilize spectral data that is usually lost in typical CNNs however helpful in most image processing tasks. We tend to evaluate the sensitivity performance of Wavelet CNNs on texture classification and image annotation. The experiments show that Wavelet CNNs can do higher accuracy in each task than existing models, whereas having significantly fewer parameters than typical CNNs.