Image classification system based on channel importance pruning and binary quantization
A binary quantization and classification system technology, applied in the direction of instruments, biological neural network models, character and pattern recognition, etc., can solve the problems of neural network memory usage and excessive calculation, reduce error fluctuations, reduce model volume, The effect of increasing the running speed
Pending Publication Date: 2021-07-27
ZHEJIANG UNIV
View PDF0 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] In order to overcome the shortcomings of the image classification method using the neural network memory occupation and excessive calculation, the present invention proposes an image classification method based on channel importance pruning and binary quantization, which reduces the model volume of the neural network and improves the running speed
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0026] The technical solution in the method of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Apparently, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.
[0027] refer to Figure 1 ~ Figure 3 , an image classification method based on channel importance pruning and binary quantization, the image classification system comprising:
[0028] The training module is used to train the weight parameters of the initial complex neural network to obtain the trained complex neural network model;
[0029] The compression module is used to repeatedly perform network pruning and restorative training based on channel importance on the trained complex neural net...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
An image classification system based on channel importance pruning and binary quantization comprises a training module used for training weight parameters of an initial complex neural network to obtain a trained complex neural network model; a compression module used for repeatedly performing network pruning and restorative training based on channel importance on the trained complex neural network model, and obtaining a preliminarily compressed neural network model on the premise of ensuring the precision, performing binary quantization on the preliminarily compressed neural network model to obtain a simplified neural network model; and a classification module used for carrying out image classification on the target image by using the compressed neural network model. The model volume of the neural network is reduced, and the operation speed is improved.
Description
technical field [0001] The invention relates to the technical field of machine learning and image classification, in particular to an image classification system based on channel importance pruning and binary quantization. Background technique [0002] Neural network is a machine learning model under deep supervised learning. It gradually extracts high-level features of images by combining low-level features. It is widely used in computer vision, including image classification, object recognition, semantic segmentation and target tracking. However, neural networks require huge storage space and computing resources, which greatly limits the application of deep neural networks on resource-constrained edge platforms. [0003] In recent years, the compression and acceleration of neural networks has gradually become a research hotspot. Among them, network pruning achieves the compression of network parameters and the improvement of inference speed by pruning out less important ch...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/082G06N3/045G06F18/241
Inventor 潘赟惠思琦朱怀宇
Owner ZHEJIANG UNIV