Loading video...

Video Failed to Load

Go Home

SAM 3 tackles a challenging problem in vision: unifying a model architecture for detection and tracking. Christoph, a researcher on SAM 3, shares how the team made it possible. 🔗 Read the SAM 3 research paper:

AIatMeta's profile picture

AI at Meta

810,922 subscribers

13,803 views • 7 months ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Collecting a high quality dataset with 4M unique phrases and 52M corresponding object masks helped SAM 3 achieve 2x the performance of baseline models. Kate, a researcher on SAM 3, explains how the data engine made this leap possible. 🔗 Read the SAM 3 research paper:
2:05

Collecting a high quality dataset with 4M unique phrases and 52M corresponding object masks helped SAM 3 achieve 2x the performance of baseline models. Kate, a researcher on SAM 3, explains how the data engine made this leap possible. 🔗 Read the SAM 3 research paper:

AI at Meta

38,684 views • 7 months ago

SAM 3D enables accurate 3D reconstruction from a single image, supporting real-world applications in editing, robotics, and interactive scene generation. Matt, a SAM 3D researcher, explains how the two-model design makes this possible for both people and complex environments. 🔗 Read the SAM 3D Objects research paper: 🔗 Read the SAM 3D Body research paper:
1:54

SAM 3D enables accurate 3D reconstruction from a single image, supporting real-world applications in editing, robotics, and interactive scene generation. Matt, a SAM 3D researcher, explains how the two-model design makes this possible for both people and complex environments. 🔗 Read the SAM 3D Objects research paper: 🔗 Read the SAM 3D Body research paper:

AI at Meta

17,858 views • 7 months ago

Meet SAM 3, a unified model that enables detection, segmentation, and tracking of objects across images and videos. SAM 3 introduces some of our most highly requested features like text and exemplar prompts to segment all objects of a target category. Learnings from SAM 3 will help power new features in Instagram Edits and Vibes, bringing advanced segmentation capabilities directly to creators. 🔗 Learn more:
1:09

Meet SAM 3, a unified model that enables detection, segmentation, and tracking of objects across images and videos. SAM 3 introduces some of our most highly requested features like text and exemplar prompts to segment all objects of a target category. Learnings from SAM 3 will help power new features in Instagram Edits and Vibes, bringing advanced segmentation capabilities directly to creators. 🔗 Learn more:

AI at Meta

189,875 views • 7 months ago

Today we’re excited to unveil a new generation of Segment Anything Models: 1️⃣ SAM 3 enables detecting, segmenting and tracking of objects across images and videos, now with short text phrases and exemplar prompts. 🔗 Learn more about SAM 3: 2️⃣ SAM 3D brings the model collection into the 3rd dimension to enable precise reconstruction of 3D objects and people from a single 2D image. 🔗 Learn more about SAM 3D: These models offer innovative capabilities and unique tools for developers and researchers to create, experiment and uplevel media workflows.
0:34

Today we’re excited to unveil a new generation of Segment Anything Models: 1️⃣ SAM 3 enables detecting, segmenting and tracking of objects across images and videos, now with short text phrases and exemplar prompts. 🔗 Learn more about SAM 3: 2️⃣ SAM 3D brings the model collection into the 3rd dimension to enable precise reconstruction of 3D objects and people from a single 2D image. 🔗 Learn more about SAM 3D: These models offer innovative capabilities and unique tools for developers and researchers to create, experiment and uplevel media workflows.

AI at Meta

1,087,568 views • 7 months ago

The fact Sam Carrick got a shift in 3 on 3 OT is hilarious. The fact Carrick scored the OT winner is even more hilarious. Sam Carrick might have just turned the page on the season. Exciting, and intense game against a good team and we won. LGR. #NYR
0:23

The fact Sam Carrick got a shift in 3 on 3 OT is hilarious. The fact Carrick scored the OT winner is even more hilarious. Sam Carrick might have just turned the page on the season. Exciting, and intense game against a good team and we won. LGR. #NYR

Sean Avery Burner

47,409 views • 1 year ago

THE PASSION. 😤 Sam Kerr puts it on a plate for Thompson to make it 3-0! #CFCW
0:27

THE PASSION. 😤 Sam Kerr puts it on a plate for Thompson to make it 3-0! #CFCW

Chelsea FC Women

33,649 views • 6 months ago

Sam Monson Sam Monson, the man who voted for Justin Herbert as MVP, shares his rankings for who comes in at #2 and #3 on his MVP list! How close was Drake Maye to securing that vital vote? 🏈
2:17

Sam Monson Sam Monson, the man who voted for Justin Herbert as MVP, shares his rankings for who comes in at #2 and #3 on his MVP list! How close was Drake Maye to securing that vital vote? 🏈

NESN

33,522 views • 4 months ago

Everyone is sleeping on Meta's SAM 3 release. But it's actually a big deal. Here's why: Companies spend millions paying humans to label images and videos frame by frame. A single autonomous driving dataset? Months of work, hundreds of annotators, millions in cost. Without labeled data, you can't train custom models. Without custom models, you're stuck with generic solutions. This is why most companies never move past pilots. SAM 3 breaks this cycle. First let's look at the evolution: SAM 1 segmented objects when you clicked on them. Revolutionary, but one object at a time. SAM 2 added video tracking with memory. Game-changing, but you still manually prompted every object. SAM 3 changes everything with text prompts. Type "yellow school bus" and it finds ALL of them in your image or video. Not just one. Every instance across thousands of frames. Now here's where people get confused: "Can't I just use GPT-5 or Gemini for this?" No, and here's why that's a terrible approach. Large multimodal LLMs are great for reasoning, but they're slow and expensive for production visual tasks. You're paying API costs per image, waiting seconds for responses, getting inconsistent results. SAM 3 runs in 30 milliseconds on a single GPU for 100+ objects. That's 100x faster, and you own the infrastructure. More importantly, SAM 3 gives you precise pixel-level masks, not descriptions. Try asking an LLM to segment every defective part on a manufacturing line in real-time. It won't work. SAM 3 does this effortlessly. The real breakthrough is their data engine. Meta built an AI-human hybrid system that's 5x faster for complex annotations. They trained SAM 3 on 4 million unique visual concepts - 50x more than existing benchmarks like LVIS. SAM 3 is trained on 4 million unique visual concepts, it handles everything: - Text-based concept search - Interactive refinement with clicks - Video tracking across frames - Zero-shot detection of new concepts The model is open source. Weights, code, and benchmarks are on GitHub. If you're building computer vision applications, this is the foundation model to evaluate. The annotation time savings alone will pay for integration costs within weeks. Find the relevant links in the next tweet!
0:27

Everyone is sleeping on Meta's SAM 3 release. But it's actually a big deal. Here's why: Companies spend millions paying humans to label images and videos frame by frame. A single autonomous driving dataset? Months of work, hundreds of annotators, millions in cost. Without labeled data, you can't train custom models. Without custom models, you're stuck with generic solutions. This is why most companies never move past pilots. SAM 3 breaks this cycle. First let's look at the evolution: SAM 1 segmented objects when you clicked on them. Revolutionary, but one object at a time. SAM 2 added video tracking with memory. Game-changing, but you still manually prompted every object. SAM 3 changes everything with text prompts. Type "yellow school bus" and it finds ALL of them in your image or video. Not just one. Every instance across thousands of frames. Now here's where people get confused: "Can't I just use GPT-5 or Gemini for this?" No, and here's why that's a terrible approach. Large multimodal LLMs are great for reasoning, but they're slow and expensive for production visual tasks. You're paying API costs per image, waiting seconds for responses, getting inconsistent results. SAM 3 runs in 30 milliseconds on a single GPU for 100+ objects. That's 100x faster, and you own the infrastructure. More importantly, SAM 3 gives you precise pixel-level masks, not descriptions. Try asking an LLM to segment every defective part on a manufacturing line in real-time. It won't work. SAM 3 does this effortlessly. The real breakthrough is their data engine. Meta built an AI-human hybrid system that's 5x faster for complex annotations. They trained SAM 3 on 4 million unique visual concepts - 50x more than existing benchmarks like LVIS. SAM 3 is trained on 4 million unique visual concepts, it handles everything: - Text-based concept search - Interactive refinement with clicks - Video tracking across frames - Zero-shot detection of new concepts The model is open source. Weights, code, and benchmarks are on GitHub. If you're building computer vision applications, this is the foundation model to evaluate. The annotation time savings alone will pay for integration costs within weeks. Find the relevant links in the next tweet!

Akshay 🚀

46,351 views • 7 months ago

The NBA’s lottery reform proposal won’t allow a team to land in the top 5 for 3 years in a row. Sam Vecenie explains why this makes no sense:
1:09

The NBA’s lottery reform proposal won’t allow a team to land in the top 5 for 3 years in a row. Sam Vecenie explains why this makes no sense:

Kevin O'Connor

61,318 views • 1 month ago

Sam Houston 7-footer Matt Dann (Matt Dann) made his presence felt in limited minutes filling up the stat sheet for Team Canada 🇨🇦 vs Switzerland 🇨🇭 👀 13 minutes played 9 Points (3/5 FG | 3/4 FT) 8 Rebounds 3 Blocks Sam Houston Men's Basketball
0:48

Sam Houston 7-footer Matt Dann (Matt Dann) made his presence felt in limited minutes filling up the stat sheet for Team Canada 🇨🇦 vs Switzerland 🇨🇭 👀 13 minutes played 9 Points (3/5 FG | 3/4 FT) 8 Rebounds 3 Blocks Sam Houston Men's Basketball

Hoops HQ

18,617 views • 11 months ago

On this date 3 years ago: The Fumble in the Jungle! Sam Hubbard’s 98 fumble return for a TD against the Ravens led the Bengals to a 24-17 playoff win. “The Hubbard Yard Dash” is one of the greatest plays in team history. Joe Burrow’s reaction: “Run faster, Sam! Go!”
0:25

On this date 3 years ago: The Fumble in the Jungle! Sam Hubbard’s 98 fumble return for a TD against the Ravens led the Bengals to a 24-17 playoff win. “The Hubbard Yard Dash” is one of the greatest plays in team history. Joe Burrow’s reaction: “Run faster, Sam! Go!”

James Rapien

39,825 views • 5 months ago

Sam Darnold and the Seahawks passing game on the last 3 drives
2:09

Sam Darnold and the Seahawks passing game on the last 3 drives

Nate Tice

27,850 views • 6 months ago

The Jets had Sam Darnold & the #2 pick They could have traded down for 3-4 firsts like the Dolphins did from #3, taken Micah Parsons, and had a huge haul of picks leftover Instead they dumped Sam for Zach Wilson SD just took his team to the Super Bowl
0:20

The Jets had Sam Darnold & the #2 pick They could have traded down for 3-4 firsts like the Dolphins did from #3, taken Micah Parsons, and had a huge haul of picks leftover Instead they dumped Sam for Zach Wilson SD just took his team to the Super Bowl

Knicks Memes

120,477 views • 5 months ago

Josh Pate Sam Houston is not an FCS team but rather a team coming off a 10-3 season and a bowl win.
0:30

Josh Pate Sam Houston is not an FCS team but rather a team coming off a 10-3 season and a bowl win.

Sports of SHSU

24,714 views • 1 year ago

SAM 3 is the missing link for vision to be as easy to use as LLMs completely changes how people will build computer vision applications you can start building with it immediately in Roboflow
0:40

SAM 3 is the missing link for vision to be as easy to use as LLMs completely changes how people will build computer vision applications you can start building with it immediately in Roboflow

Roboflow

86,644 views • 7 months ago

Sam Dekker is still a sniper -3/3 from 5 spots - 5 in a row, 1 from each spot - 90 seconds to do it Seems like a lot of NBA teams could use this (Via Sam Dekker 🎥)
1:09

Sam Dekker is still a sniper -3/3 from 5 spots - 5 in a row, 1 from each spot - 90 seconds to do it Seems like a lot of NBA teams could use this (Via Sam Dekker 🎥)

Hoop Herald

1,310,126 views • 1 year ago

Sam: "It's a team effort" Rolin: "JACOB AND SAM GO BOWLING TOGETHER" Sam: "And ice cream, and sticker play...make it up" 💀
0:27

Sam: "It's a team effort" Rolin: "JACOB AND SAM GO BOWLING TOGETHER" Sam: "And ice cream, and sticker play...make it up" 💀

PIXIE 🐰 | IWTV

130,201 views • 1 year ago

Watch Sam Altman build a custom ChatGPT using private knowledge—and share it publicly in about 3 minutes
3:31

Watch Sam Altman build a custom ChatGPT using private knowledge—and share it publicly in about 3 minutes

Dan Shipper 📧

583,906 views • 2 years ago

Sam Merrill held his matchups to 3/10 (30.0%) last night. That's the best mark on the team.
1:03

Sam Merrill held his matchups to 3/10 (30.0%) last night. That's the best mark on the team.

Mack Perry

26,570 views • 1 year ago

i’m glad i could introduce sam and jacob to the concept of a fudanshi <3
0:56

i’m glad i could introduce sam and jacob to the concept of a fudanshi <3

cuntessa

28,369 views • 17 days ago