A safety helmet detection method and system in a dynamic background

Abstract

The invention designs a safety helmet detection method and system in a dynamic background. The invention relates to the field of target detection and tracking. The method includes: labeling and training the data sets of workers and safety helmets, utilizing the characteristics of convolution neural network, inputting the features collected in the convolution layer into the classifier for training,and adding a classification layer at the end of the network to make these features enter the classifier for classification, obtaining the anchor boxes by clustering method, and predicting four coordinate values for each prediction box, drawing a prediction box for the object in the picture based on these four coordinate values, according to the predicted coordinates and categories of the predicted frame, judging whether the helmet is worn or not and the tracking of the worker is carried out. The invention can accurately identify whether a worker wears a safety helmet in a construction site or not, adopt a convolution neural network method to extract features, can better improve the identification efficiency and accuracy of the traditional method, and can better serve the shortcoming of great detection difficulty under the condition of changeable customer service background.

Description

technical field [0001] The invention relates to the field of target detection and tracking, and specifically designs a safety helmet detection method and system in a dynamic background. Background technique [0002] With the development of information technology, target detection and tracking has gradually penetrated into all aspects of people's lives, and its importance has become increasingly prominent, attracting more and more scholars and research institutions at home and abroad to participate in research in this field. At present, target detection and tracking have been widely used in video surveillance, virtual reality, human-computer interaction, planetary detection, behavior understanding and other fields. Disabled monitoring and autonomous navigation functions. [0003] Object detection is the most common problem in machine vision. It is an image segmentation based on the geometric and statistical features of the target, which combines the segmentation and recogni...

AI Technical Summary

Problems solved by technology

It is used to solve the problem of helmet detection and recognition in various places where helmets need to be worn. In order to achieve this goal, this method marks and trains the dataset of people and helmets, and uses the characteristics of the convolutional neural network. The features collected in the layer are entered into the classifier for training

Images