Loading video...

Video Failed to Load

Go Home

We are excited to present "Co-SLAM: Joint Coordinate and Sparse Parametric Encodings for Neural Real-Time SLAM" at #CVPR2023 next week. Great work by Wang Hengyi and Jingwen Wang. Check out our code and evaluation benchmark! UCL Computer Science UKRI CDT in Foundational AI

LourdesAgapito's profile picture

Lourdes Agapito

2,552 subscribers

36,863 views • 3 years ago •via X (Twitter)

#CVPR2023
LIVE

Anya Rossi• Live Now

Private livecam show

1 Comments

Frank Dellaert's profile picture
Frank Dellaert3 years ago

@Hengyi1999 @JingwenWang95 @uclcs @FAICDT1 Thanks for releasing the code. License?

Related Videos

Just met David Wang David Wang at TOKEN2049 - founder of OpenDeSci.org OpenDeSci is a next-generation science education and content platform leveraging AI to make science radically more accessible, transparent, and engaging for 8 billion people. Happy to be an advisor and investor at OpenDeSci.org - this project has real social impact in science and education.
1:36

Just met David Wang David Wang at TOKEN2049 - founder of OpenDeSci.org OpenDeSci is a next-generation science education and content platform leveraging AI to make science radically more accessible, transparent, and engaging for 8 billion people. Happy to be an advisor and investor at OpenDeSci.org - this project has real social impact in science and education.

Waqar Zaka

66,304 views • 9 months ago

Check out the Neural Graph Mapping for Dense SLAM with Efficient Loop Closure paper by Leonard Bruns, Jun Zhang, and Patric Jensfelt, visualized with Rerun. “Existing neural field-based SLAM methods typically employ a single monolithic field as their scene representation. This prevents efficient incorporation of loop closure constraints and limits scalability. To address these shortcomings, we propose a neural mapping framework which anchors lightweight neural fields to the pose graph of a sparse visual SLAM system.”
0:37

Check out the Neural Graph Mapping for Dense SLAM with Efficient Loop Closure paper by Leonard Bruns, Jun Zhang, and Patric Jensfelt, visualized with Rerun. “Existing neural field-based SLAM methods typically employ a single monolithic field as their scene representation. This prevents efficient incorporation of loop closure constraints and limits scalability. To address these shortcomings, we propose a neural mapping framework which anchors lightweight neural fields to the pose graph of a sparse visual SLAM system.”

Rerun

12,865 views • 2 years ago

Camera frame registration with little or no overlap? Check out Can Gümeli's ObjectMatch: We leverage object semantics to find indirect correspondences between frames. Great for any SLAM and SfM pose optimization! Vid: #CVPR2023 Angela Dai
0:33

Camera frame registration with little or no overlap? Check out Can Gümeli's ObjectMatch: We leverage object semantics to find indirect correspondences between frames. Great for any SLAM and SfM pose optimization! Vid: #CVPR2023 Angela Dai

Matthias Niessner

17,726 views • 3 years ago

🤩 We're so excited to share this with you: We and our friends at Cloudflare are bringing real-time voice AI to developers around the globe. 🌐 To learn how Deepgram and Cloudflare are solving real-world problems together, check out this blog:
0:29

🤩 We're so excited to share this with you: We and our friends at Cloudflare are bringing real-time voice AI to developers around the globe. 🌐 To learn how Deepgram and Cloudflare are solving real-world problems together, check out this blog:

Deepgram

19,811 views • 7 months ago

🚀 Excited to attend my first #NeurIPS2025 and present our lighting-aware SLAM method, NFL-BA! 🔗 🔦 Why this matters: Traditional Bundle Adjustment (BA) in SLAM assumes static lighting, but many real scenarios—endoscopy, search & rescue, subterranean robotics—use co-located light + camera, creating dynamic, near-field lighting that breaks these assumptions. ✨ What we introduce: Near-Field Lighting Bundle Adjustment Loss (NFL-BA) — a formulation that explicitly models near-field illumination inside the BA objective, allowing SLAM systems to jointly reason about geometry, appearance, and lighting. 📈 Results: Modeling lighting directly leads to ~38% improvement in mapping & tracking across dynamic-lighting sequences in both endoscopy and indoor scenes. 🧩 Plug-and-play: NFL-BA integrates seamlessly into existing neural rendering–based SLAM pipelines, using both implicit (NeRF-style) or explicit (3DGS) scene representations. 🌟 If you’re interested in SLAM, neural rendering, or illumination modeling, come check it out on Friday's poster session, 11-2pm, #4407.
0:38

🚀 Excited to attend my first #NeurIPS2025 and present our lighting-aware SLAM method, NFL-BA! 🔗 🔦 Why this matters: Traditional Bundle Adjustment (BA) in SLAM assumes static lighting, but many real scenarios—endoscopy, search & rescue, subterranean robotics—use co-located light + camera, creating dynamic, near-field lighting that breaks these assumptions. ✨ What we introduce: Near-Field Lighting Bundle Adjustment Loss (NFL-BA) — a formulation that explicitly models near-field illumination inside the BA objective, allowing SLAM systems to jointly reason about geometry, appearance, and lighting. 📈 Results: Modeling lighting directly leads to ~38% improvement in mapping & tracking across dynamic-lighting sequences in both endoscopy and indoor scenes. 🧩 Plug-and-play: NFL-BA integrates seamlessly into existing neural rendering–based SLAM pipelines, using both implicit (NeRF-style) or explicit (3DGS) scene representations. 🌟 If you’re interested in SLAM, neural rendering, or illumination modeling, come check it out on Friday's poster session, 11-2pm, #4407.

Roni Sengupta

21,162 views • 7 months ago

Elon Musk: “In order to solve full self-driving properly, you have to solve real-world AI. The road networks are designed to work with a biological neural net — our brains — and with vision — our eyes. So, in order to make it work with computers, you need to solve real-world AI and vision. We need cameras and silicon neural nets for self-driving to work in a system designed for eyes and biological neural nets. When you put it that way, it’s quite obvious that the only way to solve full self-driving is to solve real-world AI and sophisticated vision.”
0:56

Elon Musk: “In order to solve full self-driving properly, you have to solve real-world AI. The road networks are designed to work with a biological neural net — our brains — and with vision — our eyes. So, in order to make it work with computers, you need to solve real-world AI and vision. We need cameras and silicon neural nets for self-driving to work in a system designed for eyes and biological neural nets. When you put it that way, it’s quite obvious that the only way to solve full self-driving is to solve real-world AI and sophisticated vision.”

Elonogy

14,331 views • 25 days ago

Black in Immuno Week 2023 is almost here, we are so excited for an amazing program and can’t wait to bring our community together. Check out some highlights from last year and get registered to join us next week, Nov 9-11! #BIIW2023 #blackinimmuno
1:19

Black in Immuno Week 2023 is almost here, we are so excited for an amazing program and can’t wait to bring our community together. Check out some highlights from last year and get registered to join us next week, Nov 9-11! #BIIW2023 #blackinimmuno

BlackInImmuno

11,231 views • 2 years ago

Had a great time this week hosting 25 Grand Slam Track Racers! We celebrated their great 2024 season, bonded over our passion for this sport, and planned for an epic, game changing 2025 season! We all agreed, it’s time we focus on entertaining fans!!!
0:53

Had a great time this week hosting 25 Grand Slam Track Racers! We celebrated their great 2024 season, bonded over our passion for this sport, and planned for an epic, game changing 2025 season! We all agreed, it’s time we focus on entertaining fans!!!

Michael Johnson

39,144 views • 1 year ago

World Model Reasoning for VLM Agents (NeurIPS 2025, Score 5544) We release VAGEN to teach VLMs to build internal world models via visual state reasoning: - StateEstimation: what is the current state? - TransitionModeling: what is next? MDP → POMDP shift to handle the partial observability from visual states! 🙌Led by Kangrui Wang WilliamZhang @wzihanw Yaning Gao Linjie (Lindsey) Li Qineng Wang Hanyang Chen w4nanch1 Yiping Lu Zhengyuan Yang lijuanwang,Ranjay Krishna Jiajun Wu Fei-Fei Li Yejin Choi 👍Grateful for the joint effort of Northwestern University Computer Science Allen School Stanford AI Lab Microsoft UW–Madison Computer Sciences Siebel School of Computing and Data Science.
0:21

World Model Reasoning for VLM Agents (NeurIPS 2025, Score 5544) We release VAGEN to teach VLMs to build internal world models via visual state reasoning: - StateEstimation: what is the current state? - TransitionModeling: what is next? MDP → POMDP shift to handle the partial observability from visual states! 🙌Led by Kangrui Wang WilliamZhang @wzihanw Yaning Gao Linjie (Lindsey) Li Qineng Wang Hanyang Chen w4nanch1 Yiping Lu Zhengyuan Yang lijuanwang,Ranjay Krishna Jiajun Wu Fei-Fei Li Yejin Choi 👍Grateful for the joint effort of Northwestern University Computer Science Allen School Stanford AI Lab Microsoft UW–Madison Computer Sciences Siebel School of Computing and Data Science.

Manling Li

63,619 views • 8 months ago

I'm excited to introduce Concourse: the first AI Analyst for corporate finance teams Concourse works with finance teams to assist in day-to-day operations by taking on the most time consuming & manual work We’re also announcing our $4.7M Seed round co-led by a16z and CRV with participation from BoxGroup and Y Combinator in addition to angels and operators at leading companies like deel, Brex , Rippling, and more Check out Concourse below:
2:03

I'm excited to introduce Concourse: the first AI Analyst for corporate finance teams Concourse works with finance teams to assist in day-to-day operations by taking on the most time consuming & manual work We’re also announcing our $4.7M Seed round co-led by a16z and CRV with participation from BoxGroup and Y Combinator in addition to angels and operators at leading companies like deel, Brex , Rippling, and more Check out Concourse below:

Matthieu Hafemeister

230,634 views • 1 year ago

This seemingly obvious prediction didn't take long to become reality. MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors Contributions: • The first real-time SLAM system using the two-view 3D reconstruction prior MASt3R [20] as a foundation. • Efficient techniques for pointmap matching, tracking and local fusion, graph construction and loop closure, and second-order global optimization. • A state-of-the-art dense SLAM system capable of handling generic, time-varying camera models. Abstract: We present a real-time monocular dense SLAM system, designed from the ground up using MASt3R, a two-view 3D reconstruction and matching prior. Equipped with this strong prior, our system remains robust on in-the-wild video sequences, making no assumptions on a fixed or parametric camera model beyond a unique camera center. Key features include: - Efficient methods for pointmap matching, camera tracking, and local fusion - Graph construction and loop closure - Second-order global optimization With known calibration, a simple modification achieves state-of-the-art performance across various benchmarks. Altogether, we propose a plug-and-play monocular SLAM system capable of producing globally-consistent poses and dense geometry while operating at 15 FPS.
0:23

This seemingly obvious prediction didn't take long to become reality. MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors Contributions: • The first real-time SLAM system using the two-view 3D reconstruction prior MASt3R [20] as a foundation. • Efficient techniques for pointmap matching, tracking and local fusion, graph construction and loop closure, and second-order global optimization. • A state-of-the-art dense SLAM system capable of handling generic, time-varying camera models. Abstract: We present a real-time monocular dense SLAM system, designed from the ground up using MASt3R, a two-view 3D reconstruction and matching prior. Equipped with this strong prior, our system remains robust on in-the-wild video sequences, making no assumptions on a fixed or parametric camera model beyond a unique camera center. Key features include: - Efficient methods for pointmap matching, camera tracking, and local fusion - Graph construction and loop closure - Second-order global optimization With known calibration, a simple modification achieves state-of-the-art performance across various benchmarks. Altogether, we propose a plug-and-play monocular SLAM system capable of producing globally-consistent poses and dense geometry while operating at 15 FPS.

MrNeRF

29,935 views • 1 year ago

Excited to introduce micro1 Cortex, a contextual evaluation, visibility, and improvement platform for enterprise AI agents. Foundational models are trained for general intelligence, but enterprises need agents that perform reliably inside their unique context: workflows, policies, data environments, and edge cases. Cortex brings trust to enterprise AI by leveraging domain experts and real-world scenarios for any use case to test, diagnose, and improve how agents behave in production.
0:50

Excited to introduce micro1 Cortex, a contextual evaluation, visibility, and improvement platform for enterprise AI agents. Foundational models are trained for general intelligence, but enterprises need agents that perform reliably inside their unique context: workflows, policies, data environments, and edge cases. Cortex brings trust to enterprise AI by leveraging domain experts and real-world scenarios for any use case to test, diagnose, and improve how agents behave in production.

Ali Ansari

117,975 views • 4 months ago

ELON: TO SOLVE FULL SELF-DRIVING, YOU NEED TO SOLVE REAL-WORLD AI "What are the road networks designed to work with? They're designed to work with a biological neural net, our brains, and with vision, our eyes. And so, in order to make it work with computers, you basically need to solve real-world AI and vision. We need cameras and silicon neural nets in order to have self-driving work for a system that was designed for eyes and biological neural nets. I guess when you put it that way, it's quite obvious that the only way to solve full self-driving is to solve real-world AI and sophisticated vision." Source: Elon Musk TED Interview, April 2022
0:56

ELON: TO SOLVE FULL SELF-DRIVING, YOU NEED TO SOLVE REAL-WORLD AI "What are the road networks designed to work with? They're designed to work with a biological neural net, our brains, and with vision, our eyes. And so, in order to make it work with computers, you basically need to solve real-world AI and vision. We need cameras and silicon neural nets in order to have self-driving work for a system that was designed for eyes and biological neural nets. I guess when you put it that way, it's quite obvious that the only way to solve full self-driving is to solve real-world AI and sophisticated vision." Source: Elon Musk TED Interview, April 2022

Mario Nawfal

56,936 views • 1 year ago

Elon Musk: The only way to solve full self-driving properly is to solve real-world AI. ”What are the road networks designed to work with? They're designed to work with a biological neural net, our brains, and with vision, our eyes. And so, in order to make it work with computers, you basically need to solve real-world AI and vision. We need cameras and silicon neural nets in order to have self-driving work for a system that was designed for eyes and biological neural nets. I guess when you put it that way, it's quite obvious that the only way to solve full self-driving is to solve real-world AI and sophisticated vision.” TED Interview, April 14, 2022
0:56

Elon Musk: The only way to solve full self-driving properly is to solve real-world AI. ”What are the road networks designed to work with? They're designed to work with a biological neural net, our brains, and with vision, our eyes. And so, in order to make it work with computers, you basically need to solve real-world AI and vision. We need cameras and silicon neural nets in order to have self-driving work for a system that was designed for eyes and biological neural nets. I guess when you put it that way, it's quite obvious that the only way to solve full self-driving is to solve real-world AI and sophisticated vision.” TED Interview, April 14, 2022

ELON CLIPS

20,411 views • 1 year ago

We have been hard at work at the studio and are excited to share a reel of some of our web work. Excited for 2025!
1:10

We have been hard at work at the studio and are excited to share a reel of some of our web work. Excited for 2025!

Studio Null

27,283 views • 1 year ago

For Principles and Justice🇨🇳 —FM Wang Yi’s Week at the UN May 2026
0:43

For Principles and Justice🇨🇳 —FM Wang Yi’s Week at the UN May 2026

CHINA MFA Spokesperson 中国外交部发言人

31,828 views • 1 month ago

We are excited to share a demo of our platform. Check out our core features of creating agents, creating plugins, and interacting with them through chat. This is a work-in-progress update to give you all a sneak peek at what we’re building. There’s so much more the platform can do. We’ve laid a strong foundation, and now it’s time to refine and polish it. Stay tuned for more updates and demos over the next week. $COFFEE
3:56

We are excited to share a demo of our platform. Check out our core features of creating agents, creating plugins, and interacting with them through chat. This is a work-in-progress update to give you all a sneak peek at what we’re building. There’s so much more the platform can do. We’ve laid a strong foundation, and now it’s time to refine and polish it. Stay tuned for more updates and demos over the next week. $COFFEE

Coffee - ai/acc ☕️

10,611 views • 1 year ago

Roman Yampolskiy says we are probably in a computer simulation run by advanced AI to simulate an interesting time in history and we should try to hack out of it to take over avatar bodies in the "real world"
1:27

Roman Yampolskiy says we are probably in a computer simulation run by advanced AI to simulate an interesting time in history and we should try to hack out of it to take over avatar bodies in the "real world"

Tsarathustra

115,135 views • 1 year ago

vMAP: Vectorised Object Mapping for Neural Field SLAM, new at #CVPR2023! Each object is represented by a separate MLP, optimised in parallel via vectorised training. Full info: Dyson Robotics Lab, Imperial College, Shikun Liu, Marwan Taher, Andrew Davison.
0:24

vMAP: Vectorised Object Mapping for Neural Field SLAM, new at #CVPR2023! Each object is represented by a separate MLP, optimised in parallel via vectorised training. Full info: Dyson Robotics Lab, Imperial College, Shikun Liu, Marwan Taher, Andrew Davison.

Xin Kong

23,743 views • 3 years ago