CNN model recognition accuracy optimization method based on cross validation method

The invention relates to the technical field of artificial intelligence, and particularly discloses a CNN model recognition accuracy optimization method based on a cross validation method, and the method comprises the following steps: collecting data, and carrying out the preprocessing of the data, and obtaining a data set; carrying out K-fold division on the data set, wherein K belongs to N; performing feature construction and fusion on the data set after K-fold division, and converting the data set into a feature vector or a matrix; constructing a CNN model, inputting the converted feature vector or matrix, and optimizing the CNN model; importing the data set after K-fold division into the optimized CNN model for training and adjusting parameters; and visualizing an identification result after model training. According to the method, the obtained model recognition accuracy is more reliable, the problem of model overfitting caused by small data volume can be avoided in small data set processing, information