A video detection method and system for building digital twins

Abstract

The present application provides a video detection method and system for building digital twins. The method includes: performing entity object detection on each frame of images to be detected in continuous video image frames, and determining the entities included in each frame of images to be detected. Object category; when it is detected that the first entity object appears in any frame to be detected, the image connected region where the first entity object is located is obtained from the frame to be detected image; the image connected region is input into the line detection model , to obtain the straight line detection result of the connected region of the image; from the straight line detection result, determine the first parallel line parallel to the first right-angled side direction and the second parallel line parallel to the second right-angled side direction; according to the first parallel line and The intersection point formed by the second parallel lines in the image connected area determines the image position of the first entity object. In this way, when the present application performs video detection on a physical object with a mosaic structure, the positioning accuracy of the physical position can be improved.

Description

technical field [0001] This application relates to the technical field of architectural digital twins, in particular, to a video detection method and system for architectural digital twins. Background technique [0002] Digital twin refers to making full use of data such as physical models, sensor updates, and operation history, integrating multi-disciplinary, multi-physical quantities, multi-scale, and multi-probability simulation processes, and completing the mapping in the virtual space to reflect the entire life of the corresponding physical equipment. cycle process. Among them, architectural digital twin refers to that in the application scene of engineering construction, in order to display the spatial layout structure of the target physical building more intuitively and clearly, the engineering construction personnel usually construct a physical model that can reflect the target physical building after the target physical building is completed. The building informati...

AI Technical Summary

Problems solved by technology

[0004] In the current video detection methods, for physical objects with splicing structures such as common desks in office scenes, conventional target detection algorithms can only predict the roughly rectangular area where the desk is located in the image, and cannot accurately determine the specific position of the desk. Accurate positioning; and the problem of repeated detection may occur for large-scale targets (a whole row of desks spliced ​​out of multiple tables), which will lead to poor accuracy of entity number detection results when performing video detection on entity objects with spliced ​​structures. The problem of low accuracy and low positioning accuracy of entities, and in turn, indirectly causes distortion of the model information in the constructed / updated digital twin model

Images