Faster R-CNN-based graph model construction method

Abstract

The invention discloses a Faster R-CNN-based graph model construction method, and the method specifically comprises the steps: 1, transmitting a target image into a trained Faster R-CNN-based target detection network model, and outputting a detection frame corresponding to each target in the target image; 2, screening the target detection frames according to a set threshold value, and removing redundant target detection frames; 3, distributing a unique label to the target detection frame; 4, dividing targets of the image into primary targets and secondary targets, and then constructing a target tree for the secondary targets; 5, determining the relative position relation between the two objects; and 6, constructing a graph model of the image according to the target tree and the position relationship between the targets. According to the Faster R-CNN-based graph model construction method provided by the invention, the established graph model can effectively represent the spatial layout and semantic relation of each target in the target image.

Description

technical field [0001] The invention belongs to the technical field of computer digital image processing, and relates to a method for constructing a graph model based on Faster R-CNN. Background technique [0002] In today's rapidly developing information age, digital images, as a common and effective information carrier, have penetrated into every corner of social life, resulting in an increasing demand for image processing. [0003] In the direction of digital image processing, the target detection algorithm based on deep learning is one of the important research contents, and it has huge applications in different fields such as medical care, smart home and transportation. Compared with traditional image processing algorithms, image processing algorithms based on deep learning can extract deeper character features in complex environments, improving algorithm robustness and recognition accuracy. [0004] At present, the existing methods for constructing graph models often ...

AI Technical Summary

Problems solved by technology

[0004] At present, the existing methods for constructing graph models often use deep learning to identify objects in the target image, but the existing deep learning target recognition algorithms can only detect the category and position of the object in the graph, and cannot obtain each object in the image. The spatial relative position relationship between objects, so that the graphical model of the image cannot be well constructed

Images