METHOD AND DEVICE FOR CONVOLUTIONAL NEURAL NETWORK MODEL OPTIMIZATION, COMPUTER DEVICE, AND STORAGE MEDIUM

摘要

Disclosed are a method and device for convolutional neural network model optimization, a computer device, and a storage medium. The method relates to the artificial intelligence technology. The method comprises: dividing an initial feature matrix outputted by an input layer of a convolutional neural network model into multiple initial feature submatrices; consecutively inputting the initial feature submatrices into a convolutional layer of the convolutional neural network model to acquire feature extraction submatrices of the initial feature submatrices; superimposing the feature extraction submatrices of the initial feature submatrices to acquire an overall feature extraction matrix; and inputting the feature extraction overall matrix into the next layer of the convolutional neural network model to produce an output result.