文件列表:
图像检索中通过对抗学习的边界感知的向后兼容表示【英文版】.pdf |
下载文档 |
资源简介
>
英文标题:Boundary-aware Backward-Compatible Representation via Adversarial Learning in Image Retrieval中文摘要:介绍了在视觉检索系统中,从传统模型更新到新模型需要重新计算所有图片嵌入,所以引入兼容性训练(BCT)的方法,并提出了 Adversarial Backward-Compatible Training(AdvBCT)方法进行对比实验,表明其在兼容性和辨别性方面均优于其他 BCT 方法。英文摘要:Image retrieval plays an important role in the Internet world. Usually, thecore parts of mainstream visual retrieval systems include an online service ofthe embedding model and a large-scale vector database. For traditional modelupgrades, the old mo
加载中...
已阅读到文档的结尾了