Image recognition method and system based on lightweight convolutional neural network

Abstract

The invention discloses an image recognition method and system based on a lightweight convolutional neural network. The method comprises the steps of loading a pre-trained deep convolutional neural network model for image recognition; calculating a cutting factor of a convolution layer in the model by adopting a kernel-based sparsification method, cutting the convolution layer, further calculatinga cutting factor of each convolution kernel in the convolution layer which is not cut, and cutting the convolution kernel; for the clipped convolutional neural network model, replacing a standard convolution mode with a convolution separable packet convolution mode, and performing quantization coding on a weight coefficient matrix of a convolution kernel to obtain a lightweight convolutional neural network model; and training the lightweight model by using the image data set, and inputting a to-be-identified image into the trained lightweight convolutional neural network model for image identification. The lightweight image recognition model disclosed by the invention can be loaded to a terminal with limited computing power and storage resources, and has a relatively wide application prospect.

Description

technical field [0001] The invention relates to the field of image processing and pattern recognition, in particular to an image recognition method and system based on a lightweight convolutional neural network. Background technique [0002] With the continuous development of deep learning in the field of target recognition and detection, since AlexNet, networks such as VGG, GoogleNet, and ResNet have developed towards deeper network layers in order to seek better detection accuracy. However, many current applications, such as automatic driving, fatigue detection, and robots, are limited by integrated equipment and processing speed, and model compression research has emerged as the times require. New deep neural networks have been proposed; The number of convolution kernels and other methods are used to extract the deep-level features of the detection target; although the deep network model is superior in many problems, it is restricted by time and space in practical applica...

AI Technical Summary

Problems solved by technology

However, many current applications, such as automatic driving, fatigue detection, and robots, are limited by integrated equipment and processing speed, and model compression research has emerged as the times require. New deep neural networks have been proposed; The number of convolution kernels and other methods are used to extract the deep-level features of the detection target; although the deep network model is superior in many problems, it is restricted by time and space in practical application, and the large and deep network model has a large amount of computation. Huge, even with the help of graphics processors, it is difficult to embed and develop on devices with limited computing resources and storage resources, and it is difficult to meet the needs of many scenarios in daily life in time; high-performance computers have high production and maintenance costs and are not suitable for A large number of popularization and promotion; for example, traditional pedestrian detection equipment (such as camera monitoring head) uploads the recorded video to a remote large-scale server for data processing. Due to the influence of bandwidth and transmission delay, it is impossible to effectively detect target pedestrians in real time. Therefore, the application in some special occasions such as arresting suspects and finding lost children is relatively limited; therefore, a lightweight convolutional neural network is designed so that it can be embedded and developed on mobile terminals with limited computing and storage capabilities, making it Being able to effectively recognize image targets in real time will break through the application limitations of deep network models to a certain extent, and has broad application prospects

Images