Загрузка видео...

Не удалось загрузить видео

На главную

"Fast Foundation Stereo: When Foundation Models Meet Efficient Stereo Matching" TL;DR: distills stereo foundation models into an efficient real-time stereo matching framework, preserving high accuracy while dramatically reducing inference cost.

Almorgand's profile picture

Alexandre Morgand

2,344 subscribers

14,854 просмотров • 3 месяцев назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 1

Фото профиля Alexandre Morgand
Alexandre Morgand3 месяцев назад

, Shaurya Dewan, @BirchfieldStan @nvidia Project page: Paper: Source code:

Похожие видео

Stereo depth models usually force a tradeoff. 📌 Paper + Code Either strong zero-shot generalization or real-time speed. Fast-FoundationStereo closes that gap. The model accelerates FoundationStereo by more than 10× while keeping comparable depth quality, making real-time stereo matching practical for robotics and XR. On an RTX 3090, it produces dense metric depth maps and 3D point clouds in real time, outperforming existing real-time stereo approaches and even challenging some heavier zero-shot models. The key idea is a divide-and-conquer architecture that removes bottlenecks in common stereo matching components while preserving the strengths of the original teacher model. Training also tackles a major constraint in stereo vision: the lack of ground-truth metric depth data. The team built an automatic data curation pipeline that generates pseudo-labels from internet-scale stereo imagery using the Stereo4D dataset. Result: a real-time stereo foundation model with strong zero-shot performance and practical runtime speed. Code, pretrained weights, and pseudo-labeled stereo data will be released. Thanks for sharing, Bowen Wen! 📍Paper: Code: Project: ——- Weekly robotics and AI insights. Subscribe free:
0:25

Stereo depth models usually force a tradeoff. 📌 Paper + Code Either strong zero-shot generalization or real-time speed. Fast-FoundationStereo closes that gap. The model accelerates FoundationStereo by more than 10× while keeping comparable depth quality, making real-time stereo matching practical for robotics and XR. On an RTX 3090, it produces dense metric depth maps and 3D point clouds in real time, outperforming existing real-time stereo approaches and even challenging some heavier zero-shot models. The key idea is a divide-and-conquer architecture that removes bottlenecks in common stereo matching components while preserving the strengths of the original teacher model. Training also tackles a major constraint in stereo vision: the lack of ground-truth metric depth data. The team built an automatic data curation pipeline that generates pseudo-labels from internet-scale stereo imagery using the Stereo4D dataset. Result: a real-time stereo foundation model with strong zero-shot performance and practical runtime speed. Code, pretrained weights, and pseudo-labeled stereo data will be released. Thanks for sharing, Bowen Wen! 📍Paper: Code: Project: ——- Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

11,914 просмотров • 2 месяцев назад

MAC-VO: Metrics-Aware Covariance for Learning-based Stereo Visual Odometry TL;DR: learning-based stereo; learned metrics-aware matching uncertainty for dual purposes: selecting keypoint and weighing the residual in pose graph optimization.
0:35

MAC-VO: Metrics-Aware Covariance for Learning-based Stereo Visual Odometry TL;DR: learning-based stereo; learned metrics-aware matching uncertainty for dual purposes: selecting keypoint and weighing the residual in pose graph optimization.

Alexandre Morgand

17,367 просмотров • 1 год назад

Fast Foundation Stereo + SAM2 basically = zero shot Foundation Pose😂 No CAD model, no object image, just click the target. Run directly on my 3070 at 13fps, with a 30$ stereo camera (calibrated in 10min) Thanks Bowen Wen for his contribution to the community!
0:40

Fast Foundation Stereo + SAM2 basically = zero shot Foundation Pose😂 No CAD model, no object image, just click the target. Run directly on my 3070 at 13fps, with a 30$ stereo camera (calibrated in 10min) Thanks Bowen Wen for his contribution to the community!

Vector Wang

22,583 просмотров • 7 месяцев назад

Why can't “world models” just be sim? A real-time real2sim demo in the upcoming “ManiDreams” Foundation Stereo + SAM2 + Newton = Generalizable 3D physics prediction in a fast (15FPS) and lightweight (3070) way. No data or object mesh needed
0:21

Why can't “world models” just be sim? A real-time real2sim demo in the upcoming “ManiDreams” Foundation Stereo + SAM2 + Newton = Generalizable 3D physics prediction in a fast (15FPS) and lightweight (3070) way. No data or object mesh needed

Vector Wang

30,474 просмотров • 7 месяцев назад

We're introducing new stereo models for MusicGen. By extending the delay codebook pattern to cover tokens from both left & right channels, these models can generate stereo output with no extra computational cost vs previous models. Try MusicGen on 🤗 ➡️
0:33

We're introducing new stereo models for MusicGen. By extending the delay codebook pattern to cover tokens from both left & right channels, these models can generate stereo output with no extra computational cost vs previous models. Try MusicGen on 🤗 ➡️

AI at Meta

214,959 просмотров • 3 лет назад

60fps depth : ) these are pixels streamed over the air over wifi UDP, using fast foundation stereo i need to get end to end latency down to 20ms. inference is 10 right now. wonder if its possible (it is)
0:29

60fps depth : ) these are pixels streamed over the air over wifi UDP, using fast foundation stereo i need to get end to end latency down to 20ms. inference is 10 right now. wonder if its possible (it is)

kache

23,169 просмотров • 2 месяцев назад

มีเทียบเสียงให้ฟังคับบบ แล้วแต่คนชอบเลย แต่พึ่งเคยใช้โหมดเสียงของ mono เหมือนกัน ชอบเลยอ่ะ รู้สึกมันโอเคกว่า Stereo จริง ๆ 🥹 Stereo & Mono
0:25

มีเทียบเสียงให้ฟังคับบบ แล้วแต่คนชอบเลย แต่พึ่งเคยใช้โหมดเสียงของ mono เหมือนกัน ชอบเลยอ่ะ รู้สึกมันโอเคกว่า Stereo จริง ๆ 🥹 Stereo & Mono

𝗺𝗲𝗶𝗹𝘆 𝗲𝗶𝗲𝗶 🍒

284,492 просмотров • 11 дней назад

#willca : better in stereo
0:28

#willca : better in stereo

Juliana

54,628 просмотров • 1 год назад

Cannelle - Stereo (2026)
2:10

Cannelle - Stereo (2026)

lucía ٠࣪⭑

319,396 просмотров • 5 месяцев назад

Stereo MCs - Connected ‧ 1992
3:38

Stereo MCs - Connected ‧ 1992

70s*80s*90s*00s Throwbacks

471,875 просмотров • 2 месяцев назад

Life in Stereo 🤟
0:30

Life in Stereo 🤟

Missy K London 🔞

42,208 просмотров • 3 лет назад

Soda Stereo “Soda Stereo”, 1984. Aniversario 40 de su lanzamiento en LP y Cassette. #sodastereo #sodastereo40aniversario #cerati #ceratieterno
0:22

Soda Stereo “Soda Stereo”, 1984. Aniversario 40 de su lanzamiento en LP y Cassette. #sodastereo #sodastereo40aniversario #cerati #ceratieterno

Gustavo Cerati

68,549 просмотров • 2 лет назад

#Lukerin : Better in Stereo
0:28

#Lukerin : Better in Stereo

¡ઽα | benophie era 4eva🪁

34,946 просмотров • 8 месяцев назад

"Crema de Estrellas" Soda Stereo Teatro Gran Rex Buenos Aires Presentación Sueño Stereo #gustavocerati #ceratieterno #sodastereo #sueñostereo #ecos2026
2:20

"Crema de Estrellas" Soda Stereo Teatro Gran Rex Buenos Aires Presentación Sueño Stereo #gustavocerati #ceratieterno #sodastereo #sueñostereo #ecos2026

Cerati & Soda Stereo (Oficial Córdoba)

10,677 просмотров • 9 месяцев назад

nenhuma supera stereo love
1:18

nenhuma supera stereo love

Lucas 🦇

624,732 просмотров • 2 лет назад