Image recognition method and device based on pulsation array and medium

Abstract

The embodiment of the invention discloses an image recognition method and device based on a pulsation array, and a medium. The method comprises the steps: converting obtained image feature informationinto a one-dimensional feature vector; and converting the obtained weight matrix into a one-dimensional weight vector, and allocating a corresponding weight group to each node in the trained three-dimensional pulsation array model; performing multiplication and accumulation of the feature vectors and the weight values on the one-dimensional feature vectors in parallel by using a three-dimensionalpulsation array model to obtain a feature value corresponding to each node; enabling the feature values of different values to reflect the article category contained in the image, and determining thearticle category contained in the image according to the feature value corresponding to each node and the pre-established corresponding relationship between the feature value and the article category. According to the method, after one-dimensional conversion is carried out on the to-be-calculated image feature information and the weight value, the one-dimensional feature vector is calculated by adopting a framework of carrying out accelerated calculation by adopting the pulsation array model, so that the parallelism degree of vector calculation is fully expanded, and the classification processing efficiency of the image is effectively improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image recognition method, device and computer-readable storage medium based on a pulsation array. Background technique [0002] The current research on deep learning mainly focuses on Convolutional Neural Network (CNN). Due to different processing scenarios, the performance requirements of CNN are also different, thus developing a variety of network structures. However, the basic composition of CNN is fixed, which are input layer, convolutional layer, activation layer, pooling layer and fully connected layer. [0003] Fully connected layers (FC) play the role of "classifier" in the entire convolutional neural network. Operations such as convolutional layers, pooling layers, and activation layers map the original data to the hidden layer feature space, while the fully connected layer plays the role of mapping the learned "distributed feature representation" t...

AI Technical Summary

Problems solved by technology

[0004] At present, there are mainly two methods for calculating the fully connected layer. One is to calculate the fully connected layer according to the convolution with a convolution kernel size of 1*1, but it is necessary to replace the data of the convolution kernel, that is, the weight value very quickly. The bandwidth pressure of data transmission is very high

The second is to use the method of all or part of the parallel calculation of the elements corresponding to the feature value and the weight value. Because of the parallel calculation, an additional intermediate result processing link is required, so it cannot be continuously calculated, which reduces the use efficiency of hardware resources and affects the classification of images. efficiency

Images