paper-with-me

홈 › Papers

Augmenting Vision Language Pretraining by Learning Codebook with Visual Semantics

2022-07-31 · Xiaoyuan Guo, Jiali Duan, C. -C. Jay Kuo, Judy Wawira Gichoya, Imon Banerjee

Language modality within the vision language pretraining framework is innately discretized, endowing each word in the language vocabulary a semantic meaning. In contrast, visual modality is inherently continuous and high-dimensional, which potentially prohibits the alignment as well as fusion between vision and language modalities. We therefore propose to "discretize" the visual representation by joint learning a codebook that imbues each visual token a semantic. We then utilize these discretized visual semantics as self-supervised ground-truths for building our Masked Image Modeling objective, a counterpart of Masked Language Modeling which proves successful for language models. To optimize the codebook, we extend the formulation of VQ-VAE which gives a theoretic guarantee. Experiments validate the effectiveness of our approach across common vision-language benchmarks.

📄 PDF Abstract BibTeX arXiv:2208.00475

Code (0)

등록된 구현이 없습니다.

Tasks

Language ModelingLanguage ModellingMasked Language Modeling

Methods 이 논문이 사용한 방법론

VQ-VAE VQ-VAE is a type of variational autoencoder that uses vector quantisation to obtain a discrete latent representation. It differs from…

Similar Papers 제목 키워드 기반

VTD-CLIP: Video-to-Text Discretization via Prompting CLIP

2025-03-24 · Wencheng Zhu, Yuexin Wang, Hongxuan Li, Pengfei Zhu 외

Vision-language models bridge visual and linguistic understanding and have proven to be powerful for video recognition tasks. Existing approaches primarily rely on parameter-efficient fine-tuning of image-text pre-traine…

parameter-efficient fine-tuningVideo Recognition

ParameterNet: Parameters Are All You Need for Large-scale Visual Pretraining of Mobile Networks

2024-01-01 · CVPR 2024 1 · Kai Han, Yunhe Wang, Jianyuan Guo, Enhua Wu

The large-scale visual pretraining has significantly improve the performance of large vision models. However we observe the low FLOPs pitfall that the existing low-FLOPs models cannot benefit from large-scale pretrai…

All

MAGMA -- Multimodal Augmentation of Generative Models through Adapter-based Finetuning

2021-12-09 · Constantin Eichenberg, Sidney Black, Samuel Weinbach, Letitia Parcalabescu 외

Large-scale pretraining is fast becoming the norm in Vision-Language (VL) modeling. However, prevailing VL approaches are limited by the requirement for labeled data and the use of complex multi-step pretraining objectiv…

In-Context LearningLanguage ModelingLanguage Modelling

ParameterNet: Parameters Are All You Need

2023-06-26 · Kai Han, Yunhe Wang, Jianyuan Guo, Enhua Wu

The large-scale visual pretraining has significantly improve the performance of large vision models. However, we observe the \emph{low FLOPs pitfall} that the existing low-FLOPs models cannot benefit from large-scale pre…

All

Decoupling Vision and Language: Codebook Anchored Visual Adaptation

2026-02-23 · Jason Wu, Tianchen Zhao, Chang Liu, Jiarui Cai 외 arxiv

Large Vision-Language Models (LVLMs) use their vision encoders to translate images into representations for downstream reasoning, but the encoders often underperform in domain-specific visual tasks such as medical image …

Domain Adaptation