Multimodal Open-Vocabulary Video Classification via Pre-Trained Vision and Language Models
Research output: Working paper › Preprint › Research
Standard
Multimodal Open-Vocabulary Video Classification via Pre-Trained Vision and Language Models. / Qian, Rui; Xu, Zheng; Yang, Ming Hsuan; Belongie, Serge; Cui, Yin.
arXiv.org, 2022.Research output: Working paper › Preprint › Research
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - UNPB
T1 - Multimodal Open-Vocabulary Video Classification via Pre-Trained Vision and Language Models
AU - Qian, Rui
AU - Xu, Zheng
AU - Yang, Ming Hsuan
AU - Belongie, Serge
AU - Cui, Yin
PY - 2022
Y1 - 2022
N2 - Utilizing vision and language models (VLMs) pre-trained on large-scale image-text pairs is becoming a promising paradigm for open-vocabulary visual recognition. In this work, we extend this paradigm by leveraging motion and audio that naturally exist in video. We present \textbf{MOV}, a simple yet effective method for \textbf{M}ultimodal \textbf{O}pen-\textbf{V}ocabulary video classification. In MOV, we directly use the vision encoder from pre-trained VLMs with minimal modifications to encode video, optical flow and audio spectrogram. We design a cross-modal fusion mechanism to aggregate complimentary multimodal information. Experiments on Kinetics-700 and VGGSound show that introducing flow or audio modality brings large performance gains over the pre-trained VLM and existing methods. Specifically, MOV greatly improves the accuracy on base classes, while generalizes better on novel classes. MOV achieves state-of-the-art results on UCF and HMDB zero-shot video classification benchmarks, significantly outperforming both traditional zero-shot methods and recent methods based on VLMs. Code and models will be released.
AB - Utilizing vision and language models (VLMs) pre-trained on large-scale image-text pairs is becoming a promising paradigm for open-vocabulary visual recognition. In this work, we extend this paradigm by leveraging motion and audio that naturally exist in video. We present \textbf{MOV}, a simple yet effective method for \textbf{M}ultimodal \textbf{O}pen-\textbf{V}ocabulary video classification. In MOV, we directly use the vision encoder from pre-trained VLMs with minimal modifications to encode video, optical flow and audio spectrogram. We design a cross-modal fusion mechanism to aggregate complimentary multimodal information. Experiments on Kinetics-700 and VGGSound show that introducing flow or audio modality brings large performance gains over the pre-trained VLM and existing methods. Specifically, MOV greatly improves the accuracy on base classes, while generalizes better on novel classes. MOV achieves state-of-the-art results on UCF and HMDB zero-shot video classification benchmarks, significantly outperforming both traditional zero-shot methods and recent methods based on VLMs. Code and models will be released.
M3 - Preprint
BT - Multimodal Open-Vocabulary Video Classification via Pre-Trained Vision and Language Models
PB - arXiv.org
ER -
ID: 384580230