Abnormal sound recognition model construction method and abnormal sound detection method and system
A technology of voice recognition and construction methods, applied in speech analysis, instruments, etc., can solve the problems of privacy cannot be captured by surveillance cameras, and campus violence cannot be detected, stopped and prevented in time, so as to maintain the physical and mental safety of students and improve training. Recognition speed and accuracy, obvious effect
Pending Publication Date: 2022-07-08
JIANGSU UNIV
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0002] Due to the particularity of campus security incidents, it is difficult to be discovered and stopped in advance, and most of the victims will choose to remain silent. In high-incidence dormitories, corridors, toilets and other places, they cannot be photographed by surveillance cameras because of privacy concerns, resulting in School violence cannot be detected, stopped and prevented in time
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
[0065] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.
[0066] A method for constructing an abnormal sound recognition model, comprising the following steps:
[0067] S1. Data preparation and processing:
[0068] S1.1. When building an abnormal and dangerous sound recognition model, it is necessary to obtain a large number of audio clips from public data sets such as UrbanSound8K, TUT UrbanAcoustic Scenes 2018 challenge, etc. to form a sound data set. Some of the audio clips contain abnormal or dangerous sounds. Unusual or dangerous sounds are those that contain violent scuffles, explosions, crying, or other dangerous sounds.
[0069] Extract...
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 discloses an abnormal sound recognition model construction method and an abnormal sound detection method and system, and the method comprises the steps: constructing an abnormal dangerous sound recognition model based on a weighted non-negative matrix and a convolutional neural network, and carrying out the recognition of an abnormal sound through employing the constructed abnormal dangerous sound recognition model; and an abnormal sound detection system capable of realizing abnormal sound recognition and alarming is designed, and whether abnormal sound exists in the scene or not can be quickly and effectively detected and alarmed when the abnormal sound in the audio clip in the scene is recognized.
Description
technical field [0001] The invention relates to the technical field of smart campus and intelligent monitoring, in particular to a method and system for intelligently monitoring sounds in a campus and identifying abnormalities based on a weighted non-negative matrix and a convolutional neural network. Background technique [0002] Due to the particularity of incidents in campus security, it is difficult to detect and stop them in advance, and most of the victims choose to remain silent. In high-incidence areas such as dormitories, corridors, bathrooms and other places, they cannot be captured by surveillance cameras because of privacy concerns. School violence cannot be detected, stopped and prevented in time. Audio surveillance Compared with traditional camera-based video surveillance, traditional video surveillance is susceptible to sudden light switching, reflections and shadows, while audio sensors are insensitive to light and occlusion, are inexpensive, and are more sui...
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): G10L25/03G10L25/27G10L25/48
CPCG10L25/27G10L25/48G10L25/03
Inventor 潘雨青于浩李峰李一元蓝伟明徐乐扬
Owner JIANGSU UNIV