Web Identification Image Recognition Based on Deep Learning
Yanling Zhao, Xinchang Zhang, Xu Mei, Zhanquan Sun, Guangqi Liu, Shifeng Li · 2016
In big data era, digital information is growing rapidly. False and unlawful images influence our normal work and life, especially the exaggerated or fake propaganda of electronic commerce merchants. In this article, our purpose is to help people find out fake qualification certificate information automatically. Base on collecting and classifying web images, we apply Convolutional Neural Network (CNN) method to train a network and extract the corresponding models to recognize web identification images based on Caffe toolbox with GPU. It is saving costs on man power and material resources with high efficiency once the CNN model is trained. This model can be designed to find identification images and contrast in database. The experimental results show it is worked well in recognizing identification images with high efficiency.