Data processing method, electronic equipment and related product

Abstract

The embodiment of the invention discloses a data processing method, electronic equipment and related products, and is applied to the electronic equipment, and the method comprises the steps of inputting a first training set into a first neural network model for operation, and obtaining a first parameter model; inputting the second training set into a second neural network model for operation to obtain a second parameter model; performing operation on the second training set according to the first parameter model to obtain a second reference training set; performing operation on the first training set according to the second parameter model to obtain a first reference training set; and inputting the first reference training set into the first parameter model for operation and inputting the second reference training set into the second parameter model for operation, and taking a relatively convergent neural network model between the first reference training set and the second reference training set as a trained neural network model. By adopting the embodiment of the invention, the precision of the neural network model can be improved based on an unsupervised learning mode.

Description

technical field [0001] The present application relates to the technical field of image processing, in particular to a data processing method, electronic equipment and related products. Background technique [0002] In the past, research on face recognition has progressed slowly, because face recognition often requires large-scale or even hundreds of millions of data to support training to achieve the desired effect. Nowadays, many large-scale public datasets with manual annotations have been open sourced, which is undoubtedly It has promoted the rapid development of face recognition, and also brought an improvement in accuracy to the field of face recognition. In recent years, there have been more and more researches on face recognition, and face recognition has been widely used in monitoring scenes, community access control, mobile phones and other fields. However, in practical applications, even a model trained using a large-scale public dataset, if directly deployed to a...

AI Technical Summary

Problems solved by technology

However, in practical applications, even a model trained with a large-scale public dataset, if directly deployed to a new scene, significant scene differences usually lead to a significant drop in accuracy

In order to solve the generalization ability of the model, it is often necessary to collect, classify, and manually screen face data in new scenarios. As the data scale is getting larger and larger, the cost of manual screening will increase accordingly, which will eventually lead to manual screening. becomes unrealizable, and in turn, otherwise the model accuracy decreases

Images