Image Classification Based on Approximate Wavelet Transform and Transfer Learning on Deep Convolutional Neural Networks

In this paper a novel method has been proposed based on a combination of approximate computing, Discrete Wavelet Transform and deep neural network for image classification. In the recent trends, image classification using deep learning network comes under the limelight of the world of artificial intelligence. In the paper we have applied the approximation through bit width reduction technique through discrete wavelet transform technique for the processing of the input database. For feature extraction and classification we have developed a deep convolution neural network that is trained with that preprocesses data. The results show that the proposed model reduces the elapsed time and achieve a good rate of accuracy as compare to the Alexnet and Resnet-50 CNN models.