BriefGPT.xyz
May, 2022
CLIP-Hitchhiker的长视频检索指南
A CLIP-Hitchhiker's Guide to Long Video Retrieval
HTML
PDF
Max Bain, Arsha Nagrani, Gül Varol, Andrew Zisserman
TL;DR
本文旨在将图像-文本模型应用于长视频检索，并通过查询打分的帧嵌入的加权平均作为时间建模有效基线，提出一种在长视频检索基准测试中表现卓越的改善方法。
Abstract
Our goal in this paper is the adaptation of
image-text models
for long
video retrieval
. Recent works have demonstrated state-of-the-art performance in
→