Let RGB Be the Language of Vision

Hugging Face Daily Papers Papers
unified-vision rgb-representation image-editing zero-shot visual-language rino deep-learning

Summary

This paper introduces RINO (RGB In and RGB Out), a unified framework that represents diverse visual information (masks, depth, etc.) as RGB images and converts visual tasks into RGB-to-RGB image editing, enabling a single model to perform zero-shot transfer across tasks.

This work introduces a unified formulation for vision models, where diverse forms of visual information beyond natural images, such as masks, depth maps, and other structured visual signals, are all represented as RGB images, while general visual tasks can be converted into a common RGB-to-RGB image editing problem. In this paradigm, different types of visual information internally share the same encoding and decoding architecture and parameters as natural images, enabling a single model to transfer across tasks through a unified visual interface, in a way analogous to how language models operate over text. We refer to this formulation as RGB In and RGB Out (RINO). Built upon a generic image editing backbone without task-specific fine-tuning, RINO demonstrates robust and competitive zero-shot performance on both dense understanding tasks such as segmentation and depth estimation (where we unify outputs as RGB), and dense-conditioned generation tasks such as pose-to-image generation (where we unify inputs as RGB). We hope this study provides useful insights toward general unified vision-language systems, where diverse visual tasks can be expressed, interpreted, and solved through a shared visual language. Code is available at https://github.com/yangtiming/RINO.
Original Article
View Cached Full Text

Cached at: 07/15/26, 04:22 PM

Paper page - Let RGB Be the Language of Vision

Source: https://huggingface.co/papers/2607.12450 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

Thisworkintroducesaunifiedformulationforvisionmodels,wherediverseformsofvisualinformationbeyondnaturalimages,suchasmasks,depthmaps,andotherstructuredvisualsignals,areallrepresentedasRGBimages,whilegeneralvisualtaskscanbeconvertedintoacommonRGB-to-RGBimageeditingproblem.Inthisparadigm,differenttypesofvisualinformationinternallysharethesameencodinganddecodingarchitectureandparametersasnaturalimages,enablingasinglemodeltotransferacrosstasksthroughaunifiedvisualinterface,inawayanalogoustohowlanguagemodelsoperateovertext.WerefertothisformulationasRGBInandRGBOut(RINO).Builtuponagenericimageeditingbackbonewithouttask-specificfine-tuning,RINOdemonstratesrobustandcompetitivezero-shotperformanceonbothdenseunderstandingtaskssuchassegmentationanddepthestimation(whereweunifyoutputsasRGB),anddense-conditionedgenerationtaskssuchaspose-to-imagegeneration(whereweunifyinputsasRGB).Wehopethisstudyprovidesusefulinsightstowardgeneralunifiedvision-languagesystems,wherediversevisualtaskscanbeexpressed,interpreted,andsolvedthroughasharedvisuallanguage.Codeisavailableathttps://github.com/yangtiming/RINO.

View arXiv pageView PDFGitHub7Add to collection

Get this paper in your agent:

hf papers read 2607\.12450

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.12450 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.12450 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.12450 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Toward Robust and 3D-Aware RGB-NIR Imaging in the Dark

Hugging Face Daily Papers

This paper presents a 3D-aware neural approach for RGB-NIR low-light imaging that fuses extremely noisy RGB observations with NIR cues in 3D space, eliminating the need for clean RGB supervision and improving robustness across different noise levels.

EasyVideoR1: Easier RL for Video Understanding

Hugging Face Daily Papers

EasyVideoR1 is an efficient reinforcement learning framework for training large vision-language models on video understanding tasks, featuring offline preprocessing with tensor caching for 1.47x throughput improvement, a task-aware reward system covering 11 problem types, and evaluation across 22 video benchmarks. It also supports joint image-video training and a mixed offline-online data training paradigm.

Robostral Navigate

Hugging Face Daily Papers

Robostral Navigate is an 8B vision-language model that uses only monocular RGB images for robot navigation, achieving state-of-the-art on R2R-CE and RxR-CE benchmarks with 77.4% and 75.1% success rates, respectively.