An intelligent alarm system and method based on human body detection

Abstract

The invention provides an intelligent alarm system based on deep learning and human body detection. The method comprises the steps of constructing a human body detection database; Training Yolo-V3 network; establishing a prediction program; the camera is communicated with the GPU server; GPU server and GSM communication. According to the intelligent alarm system based on deep learning and human body detection, the diversity of application scene seasons and the complexity of human body shapes are fully considered, and a Darknet-based Yolo-is used; therefore, the human body detection speed is ensured, the human body detection accuracy is improved, and the practicability is very good.

Description

technical field [0001] The invention relates to the technical field of computer vision, and more specifically, is an intelligent alarm system and method based on human body detection. Background technique [0002] Human detection has always been a hot spot in computer vision research and applications. The main problem of human detection is to find out the person in the image or video, including the position and size. This has a strong red value, and can be combined with tracking and other technologies to be used in driverless cars, intelligent robots, passenger flow statistics, smart cities and other fields. [0003] Home security is an important issue of security issues. Among the intelligent alarm products that can be queried, there are currently more traditional products based on infrared human body detection, Bluetooth signal communication, etc. They have a common disadvantage, low accuracy, resulting in property damage Loss. For human body detection, because the huma...

AI Technical Summary

Problems solved by technology

[0003] Home anti-theft is an important issue in security issues. Among the intelligent alarm products that can be queried, there are currently relatively traditional products based on infrared human body detection and Bluetooth signal communication. Loss

For the problem of human body detection, because the human body is quite flexible, it will have various postures and shapes. At the same time, its appearance is greatly affected by clothing, posture, viewing angle, etc. In addition, it also faces factors such as occlusion and illumination, so in In current applications, the accuracy is not significant

[0004] In recent years, deep learning technology has been widely used, resulting in a series of detection networks, such as rcnn, fast-rcnn, faster-rcnn and SSD, etc. Can not be well applied to commercial products, based on this, the present invention uses Yolo-v3 network to solve the problem of human body detection, and is well embedded in the intelligent alarm system

Images