Distilling Vision-Language Models on Millions of Videos

• URL: http://arxiv.org/abs/2401.06129v2
• Date: Mon, 15 Apr 2024 21:10:37 GMT
• Title: Distilling Vision-Language Models on Millions of Videos
• Authors: Yue Zhao, Long Zhao, Xingyi Zhou, Jialin Wu, Chun-Te Chu, Hui Miao, Florian Schroff, Hartwig Adam, Ting Liu, Boqing Gong, Philipp Krähenbühl, Liangzhe Yuan,
• Abstract summary: We fine-tune a video-language model from a strong image-language baseline with synthesized instructional data. The resulting video model by video-instruction-tuning (VIIT) is then used to auto-label millions of videos to generate high-quality captions. As a side product, we generate the largest video caption dataset to date.
• Score: 62.92789440875999
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The recent advance in vision-language models is largely attributed to the abundance of image-text data. We aim to replicate this success for video-language models, but there simply is not enough human-curated video-text data available. We thus resort to fine-tuning a video-language model from a strong image-language baseline with synthesized instructional data. The resulting video model by video-instruction-tuning (VIIT) is then used to auto-label millions of videos to generate high-quality captions. We show the adapted video-language model performs well on a wide range of video-language benchmarks. For instance, it surpasses the best prior result on open-ended NExT-QA by 2.8%. Besides, our model generates detailed descriptions for previously unseen videos, which provide better textual supervision than existing methods. Experiments show that a video-language dual-encoder model contrastively trained on these auto-generated captions is 3.8% better than the strongest baseline that also leverages vision-language models. Our best model outperforms state-of-the-art methods on MSR-VTT zero-shot text-to-video retrieval by 6%. As a side product, we generate the largest video caption dataset to date.

Related papers

• Panda-70M: Captioning 70M Videos with Multiple Cross-Modality Teachers [93.65253661843145]
  We propose an automatic approach to establish a video dataset with high-quality captions. Specifically, we curate 3.8M high-resolution videos from the publicly available HD-VILA-100M dataset. We then apply multiple cross-modality teacher models to obtain captions for each video. In this way, we get 70M videos paired with high-quality text captions.
  arXiv  Detail & Related papers  (2024-02-29T18:59:50Z)
• VideoCon: Robust Video-Language Alignment via Contrast Captions [80.08882631838914]
  Video-language alignment models are not robust to semantically-plausible contrastive changes in the video captions. Our work identifies a broad spectrum of contrast misalignments, such as replacing entities, actions, and flipping event order. Our model sets new state of the art zero-shot performance in temporally-extensive video-language tasks.
  arXiv  Detail & Related papers  (2023-11-15T19:51:57Z)
• Video-Teller: Enhancing Cross-Modal Generation with Fusion and Decoupling [79.49128866877922]
  Video-Teller is a video-language foundation model that leverages multi-modal fusion and fine-grained modality alignment. Video-Teller boosts the training efficiency by utilizing frozen pretrained vision and language modules. It capitalizes on the robust linguistic capabilities of large language models, enabling the generation of both concise and elaborate video descriptions.
  arXiv  Detail & Related papers  (2023-10-08T03:35:27Z)
• MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge [35.45809761628721]
  Large scale Vision-Language (VL) models have shown tremendous success in aligning representations between visual and text modalities. We propose an unsupervised approach to tuning video data for best zero-shot action recognition performance. Our resulting models demonstrate high transferability to numerous unseen zero-shot downstream tasks.
  arXiv  Detail & Related papers  (2023-03-15T20:17:41Z)
• Towards Fast Adaptation of Pretrained Contrastive Models for Multi-channel Video-Language Retrieval [70.30052749168013]
  Multi-channel video-language retrieval require models to understand information from different channels. contrastive multimodal models are shown to be highly effective at aligning entities in images/videos and text. There is not a clear way to quickly adapt these two lines to multi-channel video-language retrieval with limited data and resources.
  arXiv  Detail & Related papers  (2022-06-05T01:43:52Z)
• Language Models with Image Descriptors are Strong Few-Shot Video-Language Learners [167.0346394848718]
  We propose VidIL, a few-shot Video-language Learner via Image and Language models. We use the image-language models to translate the video content into frame captions, object, attribute, and event phrases. We then instruct a language model, with a prompt containing a few in-context examples, to generate a target output from the composed content.
  arXiv  Detail & Related papers  (2022-05-22T05:18:27Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.