Handwritten character computer identification method

Abstract

The invention discloses a handwritten character computer identification method. The method is characterized in that the method is based on a secondary convolution neural network structure model, the model has 9 layers, the model comprises an input layer, 5 hidden layers, a full connection layer and an output layer, convolutional layers and pooling layer are staggered to form the hidden layers, the model has a Dropout layer behind the full connection layer, and the method comprises the specific steps: pre-training a filter, inputting a picture data set for training, activating a function using a Relu, outputting a characteristic vector and the like. The method is advantageous in that a convergence speed can be accelerated through the method, an over-fitting problem is solved through the method, an accumulation error is reduced through the method, and the recognition rate is improved through the method.

Description

technical field [0001] The invention belongs to the field of image recognition, and in particular relates to a computer recognition method for handwritten characters. Background technique [0002] Convolutional neural network (CNN) has attracted much attention because of its successful application in the fields of object detection, image classification, knowledge acquisition, image semantic segmentation, etc. Improving its performance is a research hotspot. In the prior art, when solving the problem of image target detection, the CPU is used to control the overall process and data scheduling of the convolutional neural network, and the GPU is used to improve the convolution calculation in the neural network unit and the calculation speed of the fully connected layer combined calculation unit, although The learning speed of the neural network has been improved, but it also increases the time cost due to the data conversion and scheduling between the CPU and GPU, and the weak ...

AI Technical Summary

Problems solved by technology

When the number of learning layers of CNN increases, their performance in solving the overfitting problem becomes worse

[0007] 3) In the training and evaluation phase, CNN adopts the gradient descent strategy, and adjusts the target gradient through the backpropagation algorithm to minimize the cumulative error of the training set. In this process, not every layer of training evaluates the cumulative error, but in a given Evaluation is performed after the interval layer, although the time to unlock will be reduced, but it will lead to an increase in cumulative error

Images