人脸识别模型训练方法及装置
By employing joint learning and adaptive margin computation, the problem of poor performance of existing face recognition models in identifying difficult samples is solved, achieving more efficient face recognition results.
CN115631522BActive Publication Date: 2026-07-17SHENZHEN XUMI YUNTU SPACE TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN XUMI YUNTU SPACE TECH CO LTD
- Filing Date
- 2022-10-24
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing face recognition models perform poorly on difficult samples, especially classic algorithms and NPCFace, which struggle to achieve ideal results.
Method used
By fusing sample pair learning and prototype learning through joint learning, and introducing the difficulty level of the samples, adaptive margin calculation and queue update mechanism are used to improve the recognition effect of difficult samples.
Benefits of technology
It significantly improves the recognition performance of face recognition models on difficult samples, and enhances the compactness of classification learning and recognition accuracy.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN115631522B_ABST
Abstract
本申请提供一种人脸识别模型训练方法及装置。该方法包括：将人脸图像样本输入到人脸识别模型的特征提取网络中，得到人脸图像样本对应的样本特征；从预设的历史样本特征队列和样本损失队列中，分别获取与人脸图像样本的类别相对应的历史样本特征和损失值；基于人脸图像样本对应的原型、损失值和历史样本特征计算得到新原型；将损失值映射为裕值，依据裕值、人脸图像样本对应的特征向量以及新原型，计算得到人脸图像样本对应的中间分类结果，将中间分类结果进行归一化处理，得到最终分类结果，以便对人脸识别模型的分类模块进行训练。本申请提升了人脸识别模型在训练过程中的人脸识别效果。
Need to check novelty before this filing date? Find Prior Art