Chinning count detection method and device based on deep convolution network

Abstract

The invention relates to a chinning count detection method and a device based on a deep convolution neural network. Firstly, a multilayer deep convolution neural network is constructed and the neural network is trained; the weight from an output layer vector element to a middle layer output vector element and the weight from the middle layer output vector element to the output layer vector element are obtained through training; according to a general error function and the relationship with an error threshold, the weights are further adjusted to obtain a good classifier; the classifier is used for analyzing an acquired video image to further detect a human body action target and action classes; the acquired class values are serially connected according to a time sequence to form an action sequence, the action sequence is compared with a standard chinning action sequence, whether the chinning of a person at the time is standard, if the sequences are matched, one is added to the chinning count. The method and the device have the advantages of strong generalization ability, strong robustness, high counting accuracy and fast counting speed.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a pull-up movement detection and counting method and device based on a deep convolutional neural network. Background technique [0002] With the continuous development of machine learning, intelligent video image recognition technology based on deep convolutional neural network has gradually become one of the research and development hotspots in the field of machine vision. The so-called "image intelligent recognition technology" is to use the deep learning algorithm to realize the key feature representation information extracted from the video image stream, and use this key feature information to provide users with valuable services. The main value is reflected in the replacement of human eyes to distinguish things and other business areas, reducing manpower input and improving work efficiency. For example, in the process of digital recognition, the recognition...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above problems, the present invention discloses a pull-up counting detection method and device based on a deep convolutional neural network. Through training the network with a large number of pull-up movement limb samples, a classifier with a high accurac

Images