The invention relates to the field of data processing, and especially relates to a convolution-like artificial neural network construction method and device. The convolution-like artificial neural network construction method comprises carrying out convolution-like artificial neural network single convolution-like layer operation on a plurality of channel feature input images; and taking an outputof a last convolution-like layer as an input of a next convolution-like layer. The convolution-like artificial neural network construction device comprises an operating module and an input/output module. The operating module is used for carrying out convolution-like artificial neural network single convolution-like layer operation on the plurality of channel feature input images. The input/outputmodule is used for taking the output of the last convolution-like layer as the input of the next convolution-like layer. By employing the convolution-like artificial neural network construction methodand device, the training time of the artificial neural network based on convolution can be shortened, and application energy consumption of the artificial neural network based on convolution can be reduced.