Image classification optimization models using the convolutional neural network ( CNN ) approach and embedded deep learning system

Deep learning has progressed rapidly in recent years and has been applied in many fields, which are the main fields of artificial intelligence. Traditional methods of machine learning most use shallow structures to deal with a limited number of samples and computational units. When the target objects have rich meanings, the performance and ability to generalize complex classification problems will be quite inadequate. The convolutional neural network (CNN), which has been developed in recent years, widely used in image processing; because it has high skills in dealing with image classification and image recognition issues and it has led to great care in many machine learning tasks and it has become a powerful and universal model of deep learning. The combination of deep learning and embedded systems has created good technical dimensions. In this paper, several useful models in the field of image classification optimization, based on convolutional neural network and embedded systems, are discussed. Since this paper focuses on usable models on the FPGA board, models known for embedded systems such as MobileNet, ResNet, ResNeXt and ShuffNet have been studied.