\n\t\n\t\t\n\t\n\t\n\t\tLinks 🔗:\n\t\n

👉 Subscribe: https://www.youtube.com/@Arxflix
👉 Twitter: https://x.com/arxflix
👉 LMNT (Partner): https://lmnt.com/

\n","updatedAt":"2024-06-09T08:27:58.762Z","author":{"_id":"6186ddf6a7717cb375090c01","avatarUrl":"/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg","fullname":"Julien BLANCHON","name":"blanchon","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":176,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.5189453363418579},"editors":["blanchon"],"editorAvatarUrls":["/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2311.04400","authors":[{"_id":"654c5a9c08031bb3deb8f326","user":{"_id":"654ca66fe1671abcbc50a6e4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/654ca66fe1671abcbc50a6e4/VqcnC17YcqC9Hk65_LxtE.jpeg","isPro":false,"fullname":"Yicong Hong","user":"YicongHong","type":"user"},"name":"Yicong Hong","status":"extracted_confirmed","statusLastChangedAt":"2023-11-09T09:33:52.791Z","hidden":false},{"_id":"654c5a9c08031bb3deb8f327","user":{"_id":"6351a882a4845f26d16a28d7","avatarUrl":"/avatars/4f8629fbf22faeb6511f4068f2828809.svg","isPro":false,"fullname":"Kai Zhang","user":"Kai-46","type":"user"},"name":"Kai Zhang","status":"claimed_verified","statusLastChangedAt":"2023-11-14T09:38:04.474Z","hidden":false},{"_id":"654c5a9c08031bb3deb8f328","user":{"_id":"642467708d97ce93878e8124","avatarUrl":"/avatars/f0e464ddb4bd790f470fc0f10275fa26.svg","isPro":false,"fullname":"Jiuxiang Gu","user":"JoshuaGu","type":"user"},"name":"Jiuxiang Gu","status":"claimed_verified","statusLastChangedAt":"2023-11-10T07:58:28.519Z","hidden":false},{"_id":"654c5a9c08031bb3deb8f329","name":"Sai Bi","hidden":false},{"_id":"654c5a9c08031bb3deb8f32a","name":"Yang Zhou","hidden":false},{"_id":"654c5a9c08031bb3deb8f32b","user":{"_id":"6393a754d31aec207924611b","avatarUrl":"/avatars/51afe808f8e8a032e34ae6d7a2d6e407.svg","isPro":false,"fullname":"Difan Liu","user":"smebliu","type":"user"},"name":"Difan Liu","status":"admin_assigned","statusLastChangedAt":"2023-11-09T09:07:30.782Z","hidden":false},{"_id":"654c5a9c08031bb3deb8f32c","name":"Feng Liu","hidden":false},{"_id":"654c5a9c08031bb3deb8f32d","name":"Kalyan Sunkavalli","hidden":false},{"_id":"654c5a9c08031bb3deb8f32e","user":{"_id":"6036eb46413a78f892e733a0","avatarUrl":"/avatars/b15925f89631323c176ecbd0eab99feb.svg","isPro":false,"fullname":"Trung Bui","user":"Trung","type":"user"},"name":"Trung Bui","status":"claimed_verified","statusLastChangedAt":"2023-11-09T09:48:12.552Z","hidden":false},{"_id":"654c5a9c08031bb3deb8f32f","name":"Hao Tan","hidden":false}],"publishedAt":"2023-11-08T00:03:52.000Z","submittedOnDailyAt":"2023-11-09T01:35:49.268Z","title":"LRM: Large Reconstruction Model for Single Image to 3D","submittedOnDailyBy":{"_id":"60f1abe7544c2adfd699860c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg","isPro":false,"fullname":"AK","user":"akhaliq","type":"user"},"summary":"We propose the first Large Reconstruction Model (LRM) that predicts the 3D\nmodel of an object from a single input image within just 5 seconds. In contrast\nto many previous methods that are trained on small-scale datasets such as\nShapeNet in a category-specific fashion, LRM adopts a highly scalable\ntransformer-based architecture with 500 million learnable parameters to\ndirectly predict a neural radiance field (NeRF) from the input image. We train\nour model in an end-to-end manner on massive multi-view data containing around\n1 million objects, including both synthetic renderings from Objaverse and real\ncaptures from MVImgNet. This combination of a high-capacity model and\nlarge-scale training data empowers our model to be highly generalizable and\nproduce high-quality 3D reconstructions from various testing inputs including\nreal-world in-the-wild captures and images from generative models. Video demos\nand interactable 3D meshes can be found on this website:\nhttps://yiconghong.me/LRM/.","upvotes":52,"discussionId":"654c5a9d08031bb3deb8f345","githubRepo":"https://github.com/3dtopia/openlrm","githubRepoAddedBy":"auto","ai_summary":"A Large Reconstruction Model using a transformer-based architecture predicts 3D neural radiance fields from single images using massive multi-view training data.","ai_keywords":["Large Reconstruction Model","transformer-based architecture","neural radiance field","ShapeNet","Objaverse","MVImgNet","end-to-end training"],"githubStars":1198},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64be5bddef8c0e42bf464d2b","avatarUrl":"/avatars/6a27b4c6a051509082d12e2222ff81fb.svg","isPro":false,"fullname":"Kevin Huang","user":"KevinHXF","type":"user"},{"_id":"6533d5abdd926ca2acc35571","avatarUrl":"/avatars/0e2ce78de1223a395e5e676756516bad.svg","isPro":false,"fullname":"Chunyi Sun","user":"Chunyi99","type":"user"},{"_id":"63e495a99db5da2dc1f679df","avatarUrl":"/avatars/e3b3f7702a5c8fd21a246272940e45b8.svg","isPro":false,"fullname":"Luming Tang","user":"lt453","type":"user"},{"_id":"62637b1bd73a349ca946f2ef","avatarUrl":"/avatars/98d64ac463419b16930a5aa745d8f5c9.svg","isPro":false,"fullname":"Taneem Ullah Jan","user":"taneemishere","type":"user"},{"_id":"6435478eef840aef2229b873","avatarUrl":"/avatars/93b6753f04fb867dcb762b9374230670.svg","isPro":false,"fullname":"haikuoxin","user":"haikuoxin","type":"user"},{"_id":"642467708d97ce93878e8124","avatarUrl":"/avatars/f0e464ddb4bd790f470fc0f10275fa26.svg","isPro":false,"fullname":"Jiuxiang Gu","user":"JoshuaGu","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":"62d5227304da46fccc445931","avatarUrl":"/avatars/108c504c92eb310f2f40ae30af756944.svg","isPro":false,"fullname":"chen","user":"bistin","type":"user"},{"_id":"644894376ffed6ece1029235","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/644894376ffed6ece1029235/8bb6OhGfVQgZiUN5iEazd.jpeg","isPro":false,"fullname":"friday","user":"kuma98","type":"user"},{"_id":"64b183300a54158d66dd3587","avatarUrl":"/avatars/55ee1f1a0bf57a9d045e739f6dbbaeed.svg","isPro":false,"fullname":"Rocl Jamez","user":"James62","type":"user"},{"_id":"654ca66fe1671abcbc50a6e4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/654ca66fe1671abcbc50a6e4/VqcnC17YcqC9Hk65_LxtE.jpeg","isPro":false,"fullname":"Yicong Hong","user":"YicongHong","type":"user"},{"_id":"6036eb46413a78f892e733a0","avatarUrl":"/avatars/b15925f89631323c176ecbd0eab99feb.svg","isPro":false,"fullname":"Trung Bui","user":"Trung","type":"user"}],"acceptLanguages":["*"],"dailyPaperRank":1}">

Papers

arxiv:2311.04400

LRM: Large Reconstruction Model for Single Image to 3D

Published on Nov 8, 2023

· Submitted by

AK on Nov 9, 2023

#1 Paper of the day

Upvote

Authors:

Yicong Hong ,

Kai Zhang ,

Jiuxiang Gu ,

Difan Liu ,

Trung Bui ,

Abstract

A Large Reconstruction Model using a transformer-based architecture predicts 3D neural radiance fields from single images using massive multi-view training data.

AI-generated summary

We propose the first Large Reconstruction Model (LRM) that predicts the 3D model of an object from a single input image within just 5 seconds. In contrast to many previous methods that are trained on small-scale datasets such as ShapeNet in a category-specific fashion, LRM adopts a highly scalable transformer-based architecture with 500 million learnable parameters to directly predict a neural radiance field (NeRF) from the input image. We train our model in an end-to-end manner on massive multi-view data containing around 1 million objects, including both synthetic renderings from Objaverse and real captures from MVImgNet. This combination of a high-capacity model and large-scale training data empowers our model to be highly generalizable and produce high-quality 3D reconstructions from various testing inputs including real-world in-the-wild captures and images from generative models. Video demos and interactable 3D meshes can be found on this website: https://yiconghong.me/LRM/.