integrated circuit

Abstract

The invention discloses an integrated circuit, which includes a data storage circuit, a weight storage circuit and an operation circuit. The data storage circuit stores the gray value data of the image within the first preset time period according to the preset control instruction, the weight storage circuit stores the data weight corresponding to the gray value data of the image within the first preset time period, and the operation circuit In the first preset time period, the gray value data and data weight are calculated, and the first data is output; the data storage circuit also stores the first data in the second preset time period according to the preset control instruction, and the weight storage circuit The data weight corresponding to the first data is also stored in the second preset time period, and the operation circuit also performs calculation on the first data and the data weight in the second preset time period, and outputs the second data. The integrated circuit in the present invention can simultaneously support the calculation of the convolutional layer and the fully connected layer, save the hardware cost of the convolutional neural network, and effectively utilize the resources of the convolutional neural network.

Description

Technical field [0001] The invention belongs to the field of image processing, and particularly relates to an integrated circuit. Background technique [0002] Convolutional Neural Networks (English full name: Convolutional Neural Networks, CNN for short) is an efficient image recognition method that has been developed in recent years and has attracted widespread attention. It has a very good performance for large-scale image processing. At present, convolutional neural networks have become one of the research hotspots in many scientific fields. In view of the fact that the convolutional neural network avoids the complicated pre-processing of the image and can directly input the original image, it has been more widely used in the field of pattern classification. [0003] Generally speaking, convolutional neural networks include convolutional layers, pooling layers and fully connected layers. In practical applications, the calculation of the convolutional layer accounts for more t...

AI Technical Summary

Problems solved by technology

[0005] The present invention provides an integrated circuit, aiming at solving the problems of high hardware cost and resource waste existing in the existing convolutional neural network

Images