Feature clustering method, database updating method, electronic equipment and storage medium

Abstract

The invention relates to the technical field of data processing, in particular to a feature clustering method and device and a database updating method and device, and the feature clustering method comprises the steps: obtaining to-be-clustered features; determining a first weight graph based on each to-be-clustered feature, nodes in the first weight graph including each to-be-clustered feature, and edges in the first weight graph being feature similarities corresponding to two nodes; inputting the first weight graph into a clustering network, and determining a primary clustering result of each to-be-clustered feature, the primary clustering result comprising an initial cluster corresponding to each to-be-clustered feature and a representative feature of each initial cluster; constructinga second weight graph based on the primary clustering result, nodes in the second weight graph comprising representative features of each initial cluster; and inputting the second weight graph into aclustering network to determine a target clustering result of each to-be-clustered feature. A hierarchical clustering framework from bottom to top is utilized to achieve the balance between the accuracy and the recall rate.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a feature clustering method, a database updating method, electronic equipment and a storage medium. Background technique [0002] Personnel recognition and vehicle recognition are an important part of security big data applications. For example, for the tracking of suspects, the pixel resolution of most surveillance videos is not enough to recognize faces. At this time, it is necessary to track the appearance characteristics of suspects , coupled with other features such as vehicles, companions, and accompanying items, the trajectory of suspects can be effectively judged. However, face recognition or vehicle recognition is an open-set problem, and people or vehicles in real scenes usually do not appear in training samples, so classification algorithms cannot be used to directly obtain people or vehicle labels through network reasoning. In contrast, feature clustering has...

AI Technical Summary

Problems solved by technology

[0004] In view of this, an embodiment of the present invention provides a feature clustering method, a database update method, electronic equipment, and a storage medium to solve the problem that it is difficult to stably achieve a balance between precision and recall caused by existing clustering algorithms

Images