Training method of convolutional neural network model
A convolutional neural network and model technology, applied in the field of convolutional neural network model training, can solve problems such as poor model accuracy and low computational efficiency
Inactive Publication Date: 2018-04-13
苏州天瞳威视电子科技有限公司
View PDF0 Cites 18 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] In view of this, the object of the present invention is to provide a training method of a convolutional neural network model, to alleviate the technical problems of poor model accuracy and low computational efficiency in the training method of the convolutional neural network model in the prior art
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
[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. the embodiment. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.
[0028] At present, the deep learning model contains a large number of parameters, which has brought a significant increase in the amount of calculations during training, directly leading to an increase in the calculation cost required for model calculations, reducing the calculation efficiency of the model, and a large number of calculations. The error rate of the calculation reduces...
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
The invention provides a training method of a convolutional neural network model and relates to the target detection technology field. Initial model training comprises steps that a training image sample is acquired by an initial model, pre-training is further carried out, and a pre-training initial model is generated; initial weight is generated according to the pre-training initial model. Convolutional neural network model training comprises steps that the initial weight is configured in the initial convolutional neural network model, a training image sample is acquired, the training learningrate of each layer in the model is set, LOSS layers are added layer by layer to the convolutional neural network model, LOSS error values are generated based on the training image sample, adjustmentfactors of the training learning rates are generated layer by layer based on the LOSS error values, training weight values are generated layer by layer according to the adjustment factors, and the training weight values are configured to the convolutional neural network model. The method is advantaged in that technical problems of low accuracy and low calculation efficiency existing in the prior art are solved, and model calculation identification accuracy and calculation efficiency are improved.
Description
technical field [0001] The invention relates to the technical field of target detection, in particular to a training method of a convolutional neural network model. Background technique [0002] With the continuous update and development of computer vision technology, object detection technology plays an important role in many fields such as intelligent transportation, image retrieval and face recognition. In recent years, the development of deep learning has become more and more popular as a more efficient tool to help us conduct research and discovery in the field of target detection. [0003] At present, deep learning has greatly surpassed traditional vision algorithms in the field of target detection. Deep learning can independently learn effective features under big data. The number and performance of learned features far exceed those of algorithm features designed by hand. [0004] In the process of implementing the present invention, the inventors have found at least...
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): G06N3/04G06N3/08
CPCG06N3/08G06N3/045
Inventor 王曦
Owner 苏州天瞳威视电子科技有限公司