Image processing method and device based on convolutional neural network, and related equipment

Abstract

The invention discloses an image processing method based on a convolutional neural network, and the method is applied to a convolution calculation unit controller, and comprises the steps: obtaining target image data; reading configuration information corresponding to the target image data; determining convolution calculation parameters of each convolution calculation unit according to the configuration information; and issuing the convolution calculation parameters to the corresponding convolution calculation unit, so that the convolution calculation unit performs convolution calculation on the target image data by using the convolution calculation parameters. The image processing method based on the convolutional neural network can be used to effectively reduce resource consumption in a convolutional neural network processing process. The invention also discloses an image processing device based on the convolutional neural network, the convolution calculation unit controller and a computer readable storage medium, which all have the above beneficial effects.

Description

technical field [0001] The present application relates to the field of artificial neural networks, in particular to an image processing method based on a convolutional neural network, and also relates to an image processing device based on a convolutional neural network, a convolutional computing unit controller, and a computer-readable storage medium. Background technique [0002] Convolutional Neural Network (CNN, Convolutional Neural Network) is a kind of artificial neural network, which is widely used in image classification, target recognition, behavior recognition, speech recognition, natural language processing and document classification and other fields. In recent years, with the growth of computer computing power and the development of CNN structure, the recognition accuracy of CNN network has been greatly improved. Therefore, heterogeneous computing devices such as GPU (Graphics Processing Unit) and FPGA (Field Programmable Gate Array, Field Programmable Gate Arra...

AI Technical Summary

Problems solved by technology

However, since inference accelerators generally adopt vectorized calculation methods, the control logic is more complicated; moreover, inference accelerators usually consist of multiple convolutional computing units to form a systolic array, and the consumption of logic resources in convolutional computing units is doubled

Images