Hugging Face Daily Papers · · 6 min read

VLM3: Vision Language Models Are Native 3D Learners

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning.","html":"<p>Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) focal length unification, 2) text-based pixel reference and 3) data mixture and scaling, are all you need for effective 3D learning. Model architecture changes, large models, heavy data augmentations, and complex losses including the regression formulation, many of which form the foundation of expert vision models, are actually not necessary conditions. As a result, we propose VLM3, a scalable method with the simplest design that enables standard VLMs to master diverse 3D tasks. VLM3 not only advances the VLM depth estimation accuracy by a large margin (0.84 -&gt; 0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning.</p>\n","updatedAt":"2026-06-01T02:07:28.509Z","author":{"_id":"6419fe794577db917f6d17fc","avatarUrl":"/avatars/ed10371a0082c5e02d32b76217200207.svg","fullname":"zhipeng cai","name":"zcai","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9132455587387085},"editors":["zcai"],"editorAvatarUrls":["/avatars/ed10371a0082c5e02d32b76217200207.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.30561","authors":[{"_id":"6a1ce909808ddbc3c7d433f6","name":"Zhipeng Cai","hidden":false},{"_id":"6a1ce909808ddbc3c7d433f7","name":"Zhuang Liu","hidden":false},{"_id":"6a1ce909808ddbc3c7d433f8","name":"Yunyang Xiong","hidden":false},{"_id":"6a1ce909808ddbc3c7d433f9","name":"Zechun Liu","hidden":false},{"_id":"6a1ce909808ddbc3c7d433fa","name":"Vikas Chandra","hidden":false},{"_id":"6a1ce909808ddbc3c7d433fb","name":"Yangyang Shi","hidden":false}],"publishedAt":"2026-05-28T00:00:00.000Z","submittedOnDailyAt":"2026-06-01T00:00:00.000Z","title":"VLM3: Vision Language Models Are Native 3D Learners","submittedOnDailyBy":{"_id":"6419fe794577db917f6d17fc","avatarUrl":"/avatars/ed10371a0082c5e02d32b76217200207.svg","isPro":false,"fullname":"zhipeng cai","user":"zcai","type":"user","name":"zcai"},"summary":"Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) focal length unification, 2) text-based pixel reference and 3) data mixture and scaling, are all you need for effective 3D learning. Model architecture changes, large models, heavy data augmentations, and complex losses including the regression formulation, many of which form the foundation of expert vision models, are actually not necessary conditions. As a result, we propose VLM3, a scalable method with the simplest design that enables standard VLMs to master diverse 3D tasks. VLM3 not only advances the VLM depth estimation accuracy by a large margin (0.84 -> 0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning.","upvotes":9,"discussionId":"6a1ce90a808ddbc3c7d433fc","githubRepo":"https://github.com/facebookresearch/VLM3","githubRepoAddedBy":"user","ai_summary":"Vision Language Models can be adapted for 3D understanding tasks through simple architectural modifications and text-based training, achieving performance comparable to specialized vision models without requiring complex designs or extensive data augmentation.","ai_keywords":["Vision Language Models","3D understanding","focal length unification","text-based pixel reference","data mixture","scaling","depth estimation","pixel correspondence","camera pose estimation","object-level 3D understanding"],"githubStars":1,"organization":{"_id":"664b1a7ccf5fd472d7d5fa1a","name":"facebook-research","fullname":"AI at Meta","avatar":"https://www.gravatar.com/avatar/7700baf1456a0ce14ca5fe9c5f27c32a?d=retro&size=100"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6419fe794577db917f6d17fc","avatarUrl":"/avatars/ed10371a0082c5e02d32b76217200207.svg","isPro":false,"fullname":"zhipeng cai","user":"zcai","type":"user"},{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"692fd35e1dbd99f126854b8d","avatarUrl":"/avatars/8e1c10584a347c88cd3109f46736a25a.svg","isPro":false,"fullname":"Yangyang Shi","user":"shiyang1983","type":"user"},{"_id":"65304b62e7535baecd85d080","avatarUrl":"/avatars/6e546c7d1414bd92c5a7c8d8c404de92.svg","isPro":false,"fullname":"Yunyang Xiong","user":"yunyangx","type":"user"},{"_id":"6407e5294edf9f5c4fd32228","avatarUrl":"/avatars/8e2d55460e9fe9c426eb552baf4b2cb0.svg","isPro":false,"fullname":"Stoney Kang","user":"sikang99","type":"user"},{"_id":"6747ede3a9c72aebe1322382","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/inILqQ05sESbYLdsEldJ_.png","isPro":false,"fullname":"Tong He","user":"tonghe90","type":"user"},{"_id":"672f5f86359d27c87963553d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-hTmvLgCD22VOVWP7Wq3L.png","isPro":false,"fullname":"Timex Peachtree","user":"TimexPeachtree","type":"user"},{"_id":"68b6973cdd7f21b758972704","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/JpeCJCu3gnRci-FLZVDTG.png","isPro":false,"fullname":"gibsonlee","user":"gibsonlee","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"664b1a7ccf5fd472d7d5fa1a","name":"facebook-research","fullname":"AI at Meta","avatar":"https://www.gravatar.com/avatar/7700baf1456a0ce14ca5fe9c5f27c32a?d=retro&size=100"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.30561.md"}">
Papers
arxiv:2605.30561

VLM3: Vision Language Models Are Native 3D Learners

Published on May 28
· Submitted by
zhipeng cai
on Jun 1
Authors:
,
,
,
,
,

Abstract

Vision Language Models can be adapted for 3D understanding tasks through simple architectural modifications and text-based training, achieving performance comparable to specialized vision models without requiring complex designs or extensive data augmentation.

AI-generated summary

Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) focal length unification, 2) text-based pixel reference and 3) data mixture and scaling, are all you need for effective 3D learning. Model architecture changes, large models, heavy data augmentations, and complex losses including the regression formulation, many of which form the foundation of expert vision models, are actually not necessary conditions. As a result, we propose VLM3, a scalable method with the simplest design that enables standard VLMs to master diverse 3D tasks. VLM3 not only advances the VLM depth estimation accuracy by a large margin (0.84 -> 0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning.

Community

Paper submitter about 9 hours ago

Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) focal length unification, 2) text-based pixel reference and 3) data mixture and scaling, are all you need for effective 3D learning. Model architecture changes, large models, heavy data augmentations, and complex losses including the regression formulation, many of which form the foundation of expert vision models, are actually not necessary conditions. As a result, we propose VLM3, a scalable method with the simplest design that enables standard VLMs to master diverse 3D tasks. VLM3 not only advances the VLM depth estimation accuracy by a large margin (0.84 -> 0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images

· Sign up or log in to comment

Get this paper in your agent:

hf papers read 2605.30561
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2605.30561 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2605.30561 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2605.30561 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

Discussion (0)

Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.

Sign in →

No comments yet. Sign in and be the first to say something.

More from Hugging Face Daily Papers