Escalator unsafe behavior real-time early warning method based on AI vision

Abstract

The invention provides an escalator unsafe behavior real-time early warning method based on AI vision in the technical field of safety monitoring. The escalator unsafe behavior real-time early warningmethod comprises the steps of: S10, acquiring an original image set of unsafe behaviors; S20, expanding the original image set by using a generative adversarial network to generate an expanded imageset; S30, annotating categories and positions of the unsafe behaviors of the image set, and generating an annotated image set; S40, creating a target detection model, and training the target detectionmodel by using the annotated image set; S50, acquiring a real-time monitoring video, decoding the real-time monitoring video, inputting the decoded real-time monitoring video into the target detection model for detection, and finding out a tracking target; S60, verifying the tracking target by using a kernel correlation filtering algorithm to generate a tracking target trajectory; and S70, generating an early warning video for early warning based on the tracking target trajectory and the real-time monitoring video. The escalator unsafe behavior real-time early warning method has the advantages that the escalator unsafe behaviors can be automatically recognized and warned in advance, and therefore escalator safety accidents are greatly reduced.

Description

technical field [0001] The invention relates to the technical field of safety monitoring, in particular to a real-time early warning method for unsafe escalator behavior based on AI vision. Background technique [0002] Escalators are one of the most typical equipment for transporting passengers in public places, and are widely used in public places such as subways, airports and shopping malls with dense traffic. In recent years, under the promotion of urbanization and aging, and under the requirements of national regulations on construction project equipment, the escalator industry will usher in a blue ocean of market. However, with the widespread use of escalators, due to the complexity of users and weak safety awareness, a single old man or child falls on the escalator, protrudes his head out of the handrail, travels backward on the escalator, and puts objects on the handrail, resulting in an accident. Safety accidents such as falling and hitting people emerge in endless...

AI Technical Summary

Problems solved by technology

However, with the widespread use of escalators, due to the complexity of the users and weak safety awareness, a single old man or child falls on the escalator, protrudes his head out of the handrail, travels retrograde on the escalator, and puts objects on the handrail, resulting in an accident. Safety accidents such as falling and smashing people emerge in endlessly, causing a large number of casualties and economic disputes. Therefore, it is imminent to strengthen the safety management of escalators in public places

[0003] However, the traditional video surveillance system only provides simple functions such as video acquisition, storage and playback, and cannot provide early warning and alarm for unsafe escalator behaviors.

In order to monitor unsafe escalator behaviors in real time and take effective measures in a timely manner, monitoring personnel need to monitor the video at all times. However, monitoring personnel are prone to fatigue, especially when faced with multi-channel surveillance video, they are often dizzy and difficult to respond to abnormal situations in a timely manner.

Images