Data annotation method and related product

Abstract

The embodiment of the invention discloses a data annotation method and a related product, which are applied to electronic equipment, and the electronic equipment is in communication connection with a manual annotation client. The method comprises the following steps: the electronic equipment obtains to-be-processed photos, determines that the to-be-processed photos form an initial photo set, extracts feature information of each to-be-processed photo in the initial photo set, determines a data feature set and a similar set through a manual annotation client according to the feature information, and determines a target annotation set through the manual annotation client according to the data feature set and the similar set; and finally data annotation is performed on each to-be-processed photo in the target annotation set according to the feature information; according to the invention, pre-labeling processing of the electronic equipment and manual classification processing are combined, labeling errors generated in the labeling process are reduced, the data labeling quality is improved, and the data labeling efficiency and accuracy are improved.

Description

technical field [0001] This application relates to the technical field of electronic equipment, in particular to a data labeling method and related products. Background technique [0002] Face recognition requires a large amount of labeled data for algorithm training. In order to complete face recognition tasks, search engines or web crawlers can be used to collect face image data from the Internet. Usually, many of the collected face images do not contain identity information. Divide the collected face image data into multiple groups of the same tagged images, where each group of tagged images corresponds to a unique subject person. It is necessary to manually merge the pictures of the same person into one set, and the pictures of different people are divided into different sets. However, some photos are of poor quality or have large differences in shooting angles. It may be difficult to judge whether the pre-labeled photos belong to the current collection when manually cle...

AI Technical Summary

Problems solved by technology

However, the quality of some photos is poor or the shooting angles are quite different. It may be difficult to judge whether the pre-labeled photos belong to the current collection when manually cleaning up the pre-labeled photos, and it may not be possible to accurately clean up the misclassified photos; when manually merging, each The collections must search for multiple similar collections according to the similarity, and judge whether they are merged by the same person one by one. This process has a large workload and is prone to omissions.

Images