Score mapping method and device, face comparison method and device, equipment and storage medium

Abstract

The invention provides a similarity score mapping method, capable of generating a mapping relation table of a standard threshold value, an actual threshold value, a negative sample error rate / positive sample passing rate by comparing a plurality of pairs of preset number of face image samples. The invention also provides a face comparison method. When the face comparison method is used for face image comparison, the similarity score compared by the face recognition model is mapped into a standard score according to the mapping relation table, and then the standard score is compared with a preset similarity threshold value to obtain a final face comparison result. The invention also provides a similarity score mapping device, a face comparison method, a face comparison device, electronic equipment and a computer readable storage medium. According to the invention, after the face recognition model is replaced, it can be guaranteed that the positive sample passing rate / the negative sample error rate is not changed without modifying the similarity threshold value, and the complexity of face image comparison operation is reduced, and the user experience is improved.

Description

technical field [0001] The present invention relates to the field of face recognition, in particular to a similarity score mapping method for face recognition, a face comparison method, a similarity score mapping device, a face comparison device, electronic equipment, and a computer-readable storage medium. Background technique [0002] In the face recognition system, for the same two photos, the similarity scores obtained by using different models are often different; using the same batch of photos, the distribution of similarity scores obtained by using different models is also different. This will cause different similarity thresholds for different models to obtain the same negative sample error rate or positive sample pass rate. If the business side wants to ensure that the negative sample error rate or positive sample pass rate remains unchanged, each time the model changes It will be necessary to modify the similarity threshold for judging whether the same person is th...

AI Technical Summary

Problems solved by technology

This will cause different similarity thresholds for different models to obtain the same negative sample error rate or positive sample pass rate. If the business side wants to ensure that the negative sample error rate or positive sample pass rate remains unchanged, each time the model changes It will be necessary to modify the similarity threshold for judging whether the same person is the same, which increases the workload when the face comparison system is upgraded

Images