A Capsule Network Image Classification Recognition Method Based on Improved Reconstruction Network

Abstract

The invention discloses a capsule network image classification and recognition method for improving the reconstructed network: S1, constructing a capsule network; S2, inputting an image training set to the capsule network, and completing image classification, recognition and calibration after the capsule network is trained and learned; S3 , input the image to be classified to the capsule network, the output vector v of the working network j The one with the largest value is the obtained recognition result; S4, the capsule network outputs the recognition result of the image to be classified; wherein the reconstructed network structure of the capsule network is a deconvolution operation. Beneficial effects: A new reconstruction network structure is proposed, the vector is restored to an image through deconvolution operation, and the error of the restored image and the original image is compared to adjust the network parameters, which reduces the amount of calculation parameters and saves space for hardware devices. More running memory.

Description

technical field [0001] The invention relates to the application of a capsule network in image classification, in particular to a capsule network image classification and recognition method for improving the reconstructed network. Background technique [0002] In recent years, convolutional neural networks have achieved rapid development in image recognition, target detection, and semantic segmentation. Convolutional neural networks are usually composed of convolutional layers, activation layers, pooling layers, and fully connected layers. The pooling layer is An important part of the convolutional neural network, typically the maximum pooling and average pooling operations, the pooling layer can reduce the size of the input feature map and reduce the amount of calculation of the model, but the pooling layer also has the loss of position information question. [0003] Aiming at the problem of loss of location information in the pooling layer of the convolutional neural netwo...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of a large amount of parameters in the capsule network, the present invention proposes a new reconstruction network structure for the existing capsule network, which restores the vector to an image through a deconvolution operation, and compares the error between the restored image and the original image to adjust Network parameters, thereby providing a capsule network image classification and recognition method that improves the reconstructed network, reduces the amount of calculation parameters, and frees up more operating memory for hardware devices

Images