Method for identifying smoking and calling in specific scene

Abstract

The invention discloses a method for identifying smoking and calling in a specific scene. The method comprises the following steps: S1, collecting a video image shot by a gun camera; S2, performing coarse positioning and recognition processing on the video image by using a human body detection algorithm; S3, if a plurality of human body targets are detected, selecting one target with the largest target area to enter S4, and if a single human body target is detected, directly skipping to S4; S4, calculating position coordinates of the central point, and sending the position coordinates of the central point to a dome camera; S5, the dome camera amplifying the human body target area to the whole picture in an automatic zooming mode, and tracking the area in real time; S6, sending video images collected by the dome camera to a target classifier; and S7, ending the steps. According to the invention, through combination of engineering means and technical means, intelligent and dead-corner-free smoking and calling behavior identification is realized, the method can be used in scenes with complex backgrounds and great changes, the effective pixel number of small-target smoking and calling is improved, the identification rate is improved, and the false alarm rate and the missing report rate are reduced.

Description

technical field [0001] The invention relates to the technical field of behavior recognition, in particular to a specific scene smoking and calling recognition method. Background technique [0002] In places such as locomotive cabs, public security interrogation rooms, coal mines, and gas stations, because of the storage of insulation materials and flammable materials, fire prevention is the top priority. Smoking or making phone calls may cause huge loss of life and property, so locomotives Smoking or making phone calls are prohibited in key places such as cabs, public security interrogation rooms, coal mines, and gas stations. Workers who are lucky may escape safety inspections. This work needs to be handed over to monitoring without dead ends. [0003] In the field of computer vision, the detection of smoking or calling is a challenging topic. Due to the small size of the target and the difficulty of detection due to light and background interference, there are currently ma...

AI Technical Summary

Problems solved by technology

[0002] In places such as locomotive cabs, public security interrogation rooms, coal mines, and gas stations, because of the storage of insulation materials and flammable materials, fire prevention is the top priority. Smoking or making phone calls may cause huge loss of life and property, so locomotives Smoking or making phone calls are prohibited in key places such as cabs, public security interrogation rooms, coal mines, and gas stations. Workers who are lucky may escape safety inspections. This work needs to be done by monitoring without dead ends.

[0003] In the field of computer vision, the detection of smoking or calling is a challenging topic. Due to the small size of the target and the difficulty of detection due to light and background interference, there are currently many literatures and patents on "smoking or calling". Most of the solutions are based on algorithms and are not combined with engineering applications, resulting in a high rate of false positives and missed negatives when smoking or making phone calls, and the scope of application is relatively limited. It can only be used in relatively closed scenes with a single background; Complex places with large scene changes

Although the existing literature uses matting to detect whether there is smoking or making a phone call, it does not increase the effective pixels in essence, especially when the image resolution is not high, the false detection is high

Images