Three-dimensional indoor model retrieval method based on deep learning

Abstract

The invention discloses a three-dimensional indoor model retrieval method based on deep learning. The method comprises the steps of establishing a model feature database; performing main body detection on the input two-dimensional image, and segmenting a main body image; performing feature vector extraction on each segmented main body image by using a feature extraction network; carrying out indoor model standard rendering graph retrieval in a model feature database by calculating the similarity of the feature vectors, and obtaining all similar indoor model standard rendering graphs; and sorting the similar indoor model standard rendering graphs according to the size sequence of the similarity, and selecting the indoor model standard rendering graph with relatively high similarity as a retrieval result to be output and displayed. The three-dimensional indoor model retrieval method based on deep learning has good distinguishing capacity for similar model individuals, meanwhile, the acquisition difficulty of a retrieval input source is reduced, and retrieval of the three-dimensional indoor model can be efficiently and accurately achieved.

Description

technical field [0001] The invention relates to a three-dimensional indoor model retrieval method, in particular to a three-dimensional indoor model retrieval method based on deep learning. Background technique [0002] With the rapid development of the field of computer graphics and the maturity of related software and hardware technologies, 3D models have been widely used in various fields, and the interior design industry is a typical representative. By using 3D design software, designers can design and modify the interior space by simply dragging, dropping and replacing the 3D interior model, which greatly improves the design efficiency of the scheme. In order to meet consumers' personalized needs for design schemes, the number of 3D indoor models has increased rapidly, and it is difficult to efficiently and accurately retrieve massive model resources using traditional keyword retrieval methods, which has become the key to restricting the development of this industry se...

AI Technical Summary

Problems solved by technology

This type of method has the advantages of low feature calculation complexity and easy indexing of the extracted features. However, since the extracted image features are mostly low-level features designed manually, high-level semantics are ignored, and the 3D model cannot be fully described. has certain limitations

Images