LOGO recognition system based on deep learning

Abstract

We used the deep learning architecture designed by ourselves to identify the logo, with good effect and accuracy. Our architecture uses four convolutional neural network architectures, two pooling structures and two fully connected neural network architecture.The characteristic of our architecture is that it is relatively simple. We can use the limited things we learn to create a program that meets our requirements.The results of the test were relatively successful. The logo recognition accuracy for our own data set can reach 95.83%.