VideoPrism: A Foundational Visual Encoder for Video Understanding

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Long, Gundavarapu, Nitesh B., Yuan, Liangzhe, Zhou, Hao, Yan, Shen, Sun, Jennifer J., Friedman, Luke, Qian, Rui, Weyand, Tobias, Zhao, Yue, Hornung, Rachel, Schroff, Florian, Yang, Ming-Hsuan, Ross, David A., Wang, Huisheng, Adam, Hartwig, Sirotenko, Mikhail, Liu, Ting, Gong, Boqing
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918047777816576
author Zhao, Long
Gundavarapu, Nitesh B.
Yuan, Liangzhe
Zhou, Hao
Yan, Shen
Sun, Jennifer J.
Friedman, Luke
Qian, Rui
Weyand, Tobias
Zhao, Yue
Hornung, Rachel
Schroff, Florian
Yang, Ming-Hsuan
Ross, David A.
Wang, Huisheng
Adam, Hartwig
Sirotenko, Mikhail
Liu, Ting
Gong, Boqing
author_facet Zhao, Long
Gundavarapu, Nitesh B.
Yuan, Liangzhe
Zhou, Hao
Yan, Shen
Sun, Jennifer J.
Friedman, Luke
Qian, Rui
Weyand, Tobias
Zhao, Yue
Hornung, Rachel
Schroff, Florian
Yang, Ming-Hsuan
Ross, David A.
Wang, Huisheng
Adam, Hartwig
Sirotenko, Mikhail
Liu, Ting
Gong, Boqing
contents We introduce VideoPrism, a general-purpose video encoder that tackles diverse video understanding tasks with a single frozen model. We pretrain VideoPrism on a heterogeneous corpus containing 36M high-quality video-caption pairs and 582M video clips with noisy parallel text (e.g., ASR transcripts). The pretraining approach improves upon masked autoencoding by global-local distillation of semantic video embeddings and a token shuffling scheme, enabling VideoPrism to focus primarily on the video modality while leveraging the invaluable text associated with videos. We extensively test VideoPrism on four broad groups of video understanding tasks, from web video question answering to CV for science, achieving state-of-the-art performance on 31 out of 33 video understanding benchmarks. Our models are released at https://github.com/google-deepmind/videoprism.
format Preprint
id arxiv_https___arxiv_org_abs_2402_13217
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle VideoPrism: A Foundational Visual Encoder for Video Understanding
Zhao, Long
Gundavarapu, Nitesh B.
Yuan, Liangzhe
Zhou, Hao
Yan, Shen
Sun, Jennifer J.
Friedman, Luke
Qian, Rui
Weyand, Tobias
Zhao, Yue
Hornung, Rachel
Schroff, Florian
Yang, Ming-Hsuan
Ross, David A.
Wang, Huisheng
Adam, Hartwig
Sirotenko, Mikhail
Liu, Ting
Gong, Boqing
Computer Vision and Pattern Recognition
Artificial Intelligence
We introduce VideoPrism, a general-purpose video encoder that tackles diverse video understanding tasks with a single frozen model. We pretrain VideoPrism on a heterogeneous corpus containing 36M high-quality video-caption pairs and 582M video clips with noisy parallel text (e.g., ASR transcripts). The pretraining approach improves upon masked autoencoding by global-local distillation of semantic video embeddings and a token shuffling scheme, enabling VideoPrism to focus primarily on the video modality while leveraging the invaluable text associated with videos. We extensively test VideoPrism on four broad groups of video understanding tasks, from web video question answering to CV for science, achieving state-of-the-art performance on 31 out of 33 video understanding benchmarks. Our models are released at https://github.com/google-deepmind/videoprism.
title VideoPrism: A Foundational Visual Encoder for Video Understanding
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2402.13217