Construction safety detection method and system based on tiny-YOLOv3

Abstract

The invention belongs to the technical field of information processing, and discloses a construction safety detection method and system based on tiny-YOLOv3. The method comprises. The method comprisesthe following steps: collecting construction personnel images in a construction site through a video monitoring camera, carrying out the manual marking of personnel contained in the images, and making a data set; training a tiny-YOLOv3 network model by using the data set; construction site image collection: a construction site image collection device acquires a construction site image and synchronously transmits the acquired image to a processor for processing; detecting the acquired images by using the trained tiny-YOLOv3 network model, and storing pictures of personnel with safety violationbehaviors (such as safety helmet wearing, smoking and the like); and the alarm information is pushed to the personnel with the safety violation behavior in a mobile phone APP mode. The method is simple in step, reasonable in design, convenient to implement, high in detection precision, good in using effect and capable of accurately detecting safety violation behaviors under the large-view condition.

Description

technical field [0001] The invention belongs to the technical field of information processing, and in particular relates to a construction safety detection method and system based on tiny-YOLOv3. Background technique [0002] At present, the closest existing technology: traditional methods mainly use color features, texture features, shape features, etc. to detect safety violations. For example, when detecting safety helmets, color features are often used for detection. First use the shape feature to detect the face area and then extract the mouth feature, and finally judge by detecting the smoke. However, because the monitoring device is affected by changes in illumination, shooting angle and scale, it is often difficult to accurately extract the above features, resulting in low detection accuracy and weak robustness. At present, there are mainly two ways to detect smoking behavior, (1) detecting the face area and then extracting mouth features to judge smoking behavior; (...

AI Technical Summary

Problems solved by technology

However, due to the complex environment of the actual construction site, and in order to monitor the entire construction site as much as possible, the monitoring device may be placed far away, so that the captured video images are in a large field of view

For this reason, when judging whether to wear a safety helmet or to use the color, shape and other characteristics of safety violations such as smoking behavior, the detection accuracy will be reduced, and it is susceptible to environmental interference

Especially when the distance is far away, affected by the image resolution, when the smoke feature is used to judge whether there is smoking behavior, the smoke feature will not be accurately extracted

[0005] To sum up, the problems existing in the existing technology are: the current image processing methods used in monitoring and preventing security violations have low detection accuracy and are susceptible to environmental interference.

However, since the monitoring device is affected by changes in illumination and shooting angles and scales, it is often difficult to accurately extract the above-mentioned features when using the above-mentioned features to detect security violations, resulting in low detection accuracy and weak robustness.

Images