English

ReVision: A Dataset and Baseline VLM for Privacy-Preserving Task-Oriented Visual Instruction Rewriting

Computation and Language 2026-01-28 v3 Artificial Intelligence Computer Vision and Pattern Recognition

Abstract

Efficient and privacy-preserving multimodal interaction is essential as AR, VR, and modern smartphones with powerful cameras become primary interfaces for human-computer communication. Existing powerful large vision-language models (VLMs) enabling multimodal interaction often rely on cloud-based processing, raising significant concerns about (1) visual privacy by transmitting sensitive vision data to servers, and (2) their limited real-time, on-device usability. This paper explores Visual Instruction Rewriting, a novel approach that transforms multimodal instructions into text-only commands, allowing seamless integration of lightweight on-device instruction rewriter VLMs (250M parameters) with existing conversational AI systems, enhancing vision data privacy. To achieve this, we present a dataset of over 39,000 examples across 14 domains and develop a compact VLM, pretrained on image captioning datasets and fine-tuned for instruction rewriting. Experimental results, evaluated through NLG metrics such as BLEU, METEOR, and ROUGE, along with semantic parsing analysis, demonstrate that even a quantized version of the model (<500MB storage footprint) can achieve effective instruction rewriting, thus enabling privacy-focused, multimodal AI applications.

Keywords

Cite

@article{arxiv.2502.14780,
  title  = {ReVision: A Dataset and Baseline VLM for Privacy-Preserving Task-Oriented Visual Instruction Rewriting},
  author = {Abhijit Mishra and Mingda Li and Hsiang Fu and Richard Noh and Minji Kim},
  journal= {arXiv preprint arXiv:2502.14780},
  year   = {2026}
}

Comments

In Proceedings of the IJCNLP-AACL 2025

R2 v1 2026-06-28T21:51:42.921Z