Residual capsule network classification model of image, classification method, equipment and storage medium
A classification model and residual technology, applied in biological neural network models, neural learning methods, character and pattern recognition, etc., can solve the problems of reduced classification accuracy and large reconstruction errors of complex data sets, so as to improve authenticity and benefit The effect of network parameter optimization and improvement of classification accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0058] A residual capsule network classification model for images, such as figure 1 As shown, it includes a residual extraction module, a capsule module and a reconstruction module connected in sequence; the residual extraction module is used to extract the deep information of the image to be classified into the capsule layer; the capsule module is used to achieve classification. The capsule layer realizes the conversion of scalar neurons to vector neurons, uses the routing algorithm to realize the iterative update of the parameters between the main capsule and the digital capsule, and uses the digital capsule layer to identify the category of the feature; the reconstruction module is used to reconstruct the image, for digital The recognition result of the capsule layer uses deconvolution to reconstruct the image to obtain the classified image.
[0059] Among them, the residual extraction module is composed of three sequentially connected residual extraction blocks (REB); the ...
Embodiment 2
[0068] A classification method for an image residual capsule network classification model, such as Figure 1-2 shown, follow the steps below:
[0069] Step S1, inputting the image to be classified into the residual extraction module of the image residual capsule network classification model;
[0070] Step S2, the residual extraction module first uses a layer of 3×3 convolution, batch normalization and beta-mish activation layer to extract the primary features of the image to be classified, and then passes a layer of 1×1 convolution, batch normalization and The beta-mish activation layer further extracts the input features, and uses three sequentially connected residual extraction blocks (REB) to extract the deep features of the image to be classified, and obtains the deep information used to extract the image to be classified into the capsule layer; the primary features and The deep information is both independent and correlated, and each image must contain primary features (...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More