A hierarchical semantic embedding model for fine object recognition and implementation method thereof

Abstract

The invention discloses a hierarchical semantic embedding model for fine recognition of an object and an implementation method thereof. The hierarchical semantic embedding model comprises a backbone network, which is used for extracting shallow features of an input image and outputting the features to each branch network in the form of a feature map; several branch networks, which are used for further extracting deep-seated features from the image shallow feature map output by the backbone network, so that the output characteristic map is suitable for the identification task of the corresponding hierarchy of the branch network, by introducing the semantic knowledge embedding mechanism, the invention realizes the guidance of the upper semantic knowledge to the feature learning of the lowerbranch network. The invention solves the problem that the additional information labeling cost is high in the object refinement identification technical scheme which relies on the additional information to guide learning.

Description

technical field [0001] The invention relates to the technical field of object fine recognition, in particular to a hierarchical semantic embedding (Hierarchical Semantic Embedding, HSE) model and a training method thereof for fine object recognition. Background technique [0002] In recent years, the transformation of deep visual computing has ignited the demand for visual understanding and analysis technology in various fields, such as the urgent need for online and accurate retrieval of clothing pictures for e-commerce, the urgent need for accurate matching of vehicles involved in the security industry, and the urgent need for fine identification of wild animals and plants in the agricultural, forestry and environmental protection circles, etc. Wait. These requirements often require the recognition algorithm to be able to distinguish the subordinate categories of a certain basic category in detail, and this technology is usually called the fine recognition of objects. [...

AI Technical Summary

Problems solved by technology

[0010] In order to overcome the deficiencies in the above-mentioned prior art, the purpose of the present invention is to provide a hierarchical semantic embedding model for fine object recognition and its implementation method, so as to solve the additional problems in the technical solution of fine object recognition that relies on additional information to guide learning. The problem of high cost of information labeling

Images