Detection method of illegal pouring objects in concrete pouring based on artificial intelligence video analysis

Abstract

The invention provides a method for detecting illegal pouring objects in concrete pouring based on artificial intelligence video analysis. The method includes: performing event segmentation and judgment on the video stream uploaded in real time through the CoViAR-Mobile-V3 algorithm to obtain the video stream of the pouring link; using SSD The detection algorithm detects and locates the area to be poured, determines its boundary, detects suspected foreign objects, and outputs the location area of ​​​​the sub-area; for the relevant context video stream of the sub-area, detects and locates whether there are people or objects in the pouring area, and obtains the suspected area window; Behavior recognition is performed on the sub-video stream in this area. If there is pouring or object throwing behavior, whether it is concrete or not, it will be regarded as an illegal behavior, an early warning will be issued, and then structured storage will be carried out. The present invention detects, supervises and manages illegal pouring objects in real time through intelligent video analysis, gives early warning to events that do not conform to or are suspected of conforming to specifications, and uploads them to a server for storage, thereby largely avoiding the related defects of manual on-site supervision and management.

Description

technical field [0001] The invention relates to the technical field of construction engineering, in particular to a method for detecting illegally poured objects in concrete pouring based on artificial intelligence video analysis. Background technique [0002] Concrete pouring quality management is an important part of the supervision work at the site. In the process of pouring quality management, illegal pouring behaviors often bring great risks of safety accidents. At present, in the process of manual supervision, professional supervisors conduct real-time on-site supervision of pouring behavior through manual observation on the spot to strictly prevent illegal pouring materials from entering the final hole. During the supervision process, due to the lack of effective supervision methods such as images, The professional quality of the supervisors is poor, or due to the exchange of interests, it is easy to cause ineffective actions, little actions, malicious omissions, etc....

AI Technical Summary

Problems solved by technology

[0003] Under the existing manual on-site supervision and management mode, its shortcomings are more obvious. The specific developments are as follows: the quality of side station supervision and management has the following high-risk problems: the quality of side station supervisors is uneven, and it is impossible to fully grasp the side station supervision. The key points of the side station can easily lead to the loss of control of the side station process; the supervisors have great discretion, and the lack of image data support for the side station process and side station management results can easily lead to the supervision accepting benefits and relaxing management; during the side station supervision process, due to the side station The supervisors lack restraint, and it often happens that only the side stations do nothing or do little; the supervision records of the side stations are not serious and have no traceability; during the side station process, there are safety and quality hazards or violations of mandatory standards for engineering construction. Behavior, stay in rectification suggestions or verbal notification, do not track and close the rectification situation

Images