Method for fast search and effective storage of similar faces

Abstract

The invention relates to a method for rapid search and effective storage of similar faces, including the following steps: creating a face information list; searching for similar faces: sending the face information in the face entry and the captured face pictures into the SDK together Perform face comparison; send the face recognition result together with the captured face image to the step of updating the face information list; assemble a new face entry according to the captured face image, and update the face information list. The method for fast search and effective storage of similar faces of the present invention uses the method of linking the captured face pictures with a chain structure of similar faces and a face database to perform a fast search for similar faces, and uses the MongoDB database to store the associated similar faces. , which reduces the amount of face search operations and improves the search speed. The capacity of the MongoDB database for storing faces changes dynamically according to the size of the actual hard disk. Good practical value.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence and data retrieval, and in particular relates to a method for quickly searching and effectively storing similar faces. Background technique [0002] At present, face recognition systems are widely used in various industries such as safe cities, transportation hubs, smart medical care, commercial chains, border inspections, etc. This has also led to more and more companies joining the security field, but the threshold of face recognition technology itself is relatively high. , although there are domestic companies such as SenseTime, Megvii, Yitu, and Yuncong that specialize in face recognition algorithm research, most companies still cannot invest huge R&D costs in algorithm research in the short term, so they prefer Use the third-party face recognition SDK to do customized development for industry applications. [0003] Although the face recognition SDKs launched by various algori...

AI Technical Summary

Problems solved by technology

[0002] At present, face recognition systems are widely used in various industries such as safe cities, transportation hubs, smart medical care, commercial chains, border inspections, etc. This has also led to more and more companies joining the security field, but the threshold of face recognition technology itself is relatively high. , although there are domestic companies such as SenseTime, Megvii, Yitu, and Yuncong that specialize in face recognition algorithm research, most companies still cannot invest huge R&D costs in algorithm research in the short term, so they prefer Use third-party face recognition SDK to do customized development for industry applications

[0003] Although the face recognition SDKs launched by various algorithm manufacturers have different recognition performances, they all have one thing in common: the capacity of the face database is limited. The larger the capacity of the face database, the higher the price of the SDK.

[0004] This mode leads to poor capacity scalability of the face database, which is not suitable for some application scenarios, such as face pictures captured by cameras in real time and sent to the SDK to complete the face storage operation.

[0005] Before entering the database, the SDK will perform face recognition on the face picture. Faces that reach the recognition threshold will be assigned the same personID and identified as similar faces. It is the same person and the person’s trajectory is analyzed, so the face data collected by the camera in real time needs to be continuously stored in the database, which will cause the face database to easily reach the upper limit. Once the upper limit is reached, the subsequent captured faces will not be valid. Identification and Inbound Storage

[0006] In addition, the traditional face recognition method needs to compare the captured face with each face in the face database to search for the face with the highest similarity. Once the face database is large, similar faces Not only will the efficiency of the search be reduced, but the failed face pictures will either be discarded or simply stored in the database, which will not be helpful for later similar face searches

Images