正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

🚨Crowded scenes are notoriously difficult for #ComputerVision models. In our new work lead by Zhou Mu & Lucas Stoffl we developed a novel approach called BUCTD that is state-of-the-art on crowded 🐒🐠&🕺human pose estimation benchmarks!🔥 #CVPR2023 workshop

Mackenzie Weygandt Mathis, PhD

22,785 subscribers

67,217 次观看 • 3 年前 •via X (Twitter)

#ComputerVision #CVPR2023

Anya Rossi• Live Now

Private livecam show

6 条评论

Mackenzie Mathis, PhD 的头像

Mackenzie Mathis, PhD3 年前

If you’re at #CVPR2023 go check out @LStoffl presenting our accepted short form paper version of this TODAY! And sadly co-first author @zhoumu53 couldn’t get a CAN visa 🤬, so go support her work even if she can’t be there too!

Dan Biderman 的头像

Dan Biderman3 年前

@zhoumu53 @LStoffl Very cool model idea, and results! congrats.

Niamato Inc 的头像

Niamato Inc3 年前

@zhoumu53 @LStoffl Remarkable

Elvira De Leonibus 的头像

Elvira De Leonibus3 年前

@zhoumu53 @LStoffl @DilettaCavezza

Rui-Di Wang 的头像

Rui-Di Wang3 年前

@zhoumu53 @LStoffl So cool!

Felipe Arruda Moura 的头像

Felipe Arruda Moura3 年前

@zhoumu53 @LStoffl Fantastic! Have you already shared the github of that? Thanks!!

相关视频

Meta presents Sapiens Foundation for Human Vision Models discuss: We present Sapiens, a family of models for four fundamental human-centric vision tasks - 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability - model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error.

Meta presents Sapiens Foundation for Human Vision Models discuss: We present Sapiens, a family of models for four fundamental human-centric vision tasks - 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability - model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error.

AK

151,614 次观看 • 2 年前

Announcing Gaia, a major update to our AI phone agents. Gaia sets a new state of the art: its latency beats the competition by over 500ms in our benchmarks. This is the closest that phone agents have ever gotten to human calls. Give it a call at the number below!

Announcing Gaia, a major update to our AI phone agents. Gaia sets a new state of the art: its latency beats the competition by over 500ms in our benchmarks. This is the closest that phone agents have ever gotten to human calls. Give it a call at the number below!

Flo Crivello

518,625 次观看 • 10 个月前

This was September 23rd. “The work that we are doing to ensure that there is peace in the Middle East is a step by step approach and one of those steps is to recognize the state of Palestine” So much for that “step by step approach” Trump got the deal done without UN marching orders. Our Government is brutal.

This was September 23rd. “The work that we are doing to ensure that there is peace in the Middle East is a step by step approach and one of those steps is to recognize the state of Palestine” So much for that “step by step approach” Trump got the deal done without UN marching orders. Our Government is brutal.

Ryan Gerritsen🇨🇦🇳🇱

31,877 次观看 • 10 个月前

🚨BREAKING🚨 It is now being reported that DOZENS of Americans are being held hostage by the Iranian-backed Hamas terrorists in Gaza. Iran is NOT our friend. This is TERRORISM. We pray for the safe return of our countrymen & we pray for safety & security of our ally Israel.

🚨BREAKING🚨 It is now being reported that DOZENS of Americans are being held hostage by the Iranian-backed Hamas terrorists in Gaza. Iran is NOT our friend. This is TERRORISM. We pray for the safe return of our countrymen & we pray for safety & security of our ally Israel.

Congressman Byron Donalds

462,769 次观看 • 2 年前

🔥 BREAKING 🔥 The Australian Government have imported so many dangerous foreign criminals & Islamic terrorists that now we need a plan for public crowded places so that they don’t kill us. Our Government brings them here, repeatedly gives them bail even after stabbing & killing people so it is only a matter of time before we have a serious mass casualty event in our major cities. “The counter terrorism advice focuses on crowded venues, like airports and shopping centres, at risk of being targeted by extremists” -7 News

🔥 BREAKING 🔥 The Australian Government have imported so many dangerous foreign criminals & Islamic terrorists that now we need a plan for public crowded places so that they don’t kill us. Our Government brings them here, repeatedly gives them bail even after stabbing & killing people so it is only a matter of time before we have a serious mass casualty event in our major cities. “The counter terrorism advice focuses on crowded venues, like airports and shopping centres, at risk of being targeted by extremists” -7 News

Lozzy B 🇦🇺𝕏

27,556 次观看 • 10 个月前

I thank the Secretary of State for his continued support of the new The Hillingdon Hospitals Hospital – enabling works are already underway as we get ready to build a superb state-of-the-art facility that will transform healthcare in Uxbridge & South Ruislip.

I thank the Secretary of State for his continued support of the new The Hillingdon Hospitals Hospital – enabling works are already underway as we get ready to build a superb state-of-the-art facility that will transform healthcare in Uxbridge & South Ruislip.

Boris Johnson

332,746 次观看 • 3 年前

This is so refreshing from Jonathan Reynolds our new Secretary of State for Business & Trade. A Business secretary actually answering questions honestly, thoughtfully & without rhetoric. A closer relationship with the EU is needed & at last we are now having that conversation

This is so refreshing from Jonathan Reynolds our new Secretary of State for Business & Trade. A Business secretary actually answering questions honestly, thoughtfully & without rhetoric. A closer relationship with the EU is needed & at last we are now having that conversation

Peter Stefanovic

369,465 次观看 • 2 年前

Our vision is for AI that uses world models to adapt in new and dynamic environments and efficiently learn new skills. We’re sharing V-JEPA 2, a new world model with state-of-the-art performance in visual understanding and prediction. V-JEPA 2 is a 1.2 billion-parameter model, trained on video, that can enable zero-shot planning in robots—allowing them to plan and execute tasks in unfamiliar environments. Learn more about V-JEPA 2 ➡️ As we continue working toward our goal of achieving advanced machine intelligence (AMI), we’re also releasing three new benchmarks for evaluating how well existing models can reason about the physical world from video. Learn more and download the new benchmarks ➡️

Our vision is for AI that uses world models to adapt in new and dynamic environments and efficiently learn new skills. We’re sharing V-JEPA 2, a new world model with state-of-the-art performance in visual understanding and prediction. V-JEPA 2 is a 1.2 billion-parameter model, trained on video, that can enable zero-shot planning in robots—allowing them to plan and execute tasks in unfamiliar environments. Learn more about V-JEPA 2 ➡️ As we continue working toward our goal of achieving advanced machine intelligence (AMI), we’re also releasing three new benchmarks for evaluating how well existing models can reason about the physical world from video. Learn more and download the new benchmarks ➡️

AI at Meta

310,340 次观看 • 1 年前

Gen.Muhoozi Kainerugaba: With a new group of leaders in cabinet & parliament,we are optimistic that Uganda is on track to achieve it's target of $500b economy by 2040 & as for UPDF we shall perform our duty to enable this.we make this pledge as gallant servants of our country.

Gen.Muhoozi Kainerugaba: With a new group of leaders in cabinet & parliament,we are optimistic that Uganda is on track to achieve it's target of $500b economy by 2040 & as for UPDF we shall perform our duty to enable this.we make this pledge as gallant servants of our country.

Robert2026

19,902 次观看 • 1 个月前

📢 OneCanvas: 3D Scene Understanding via Panoramic Reprojection We extract features from video frames and reproject them into one occlusion-free view of the whole scene that a 2D VLM reads just like a normal image. We can center this view on any viewpoint, including an agent's own pose for situated reasoning. The same projection lets us create spatial training tasks with no human annotation, solvable only by reasoning over the 3D positions of real object features placed on an otherwise empty canvas. The result is a stock 2D VLM that reasons in 3D, setting a new state of the art across spatial benchmarks at far less compute. 🌐 ▶️ Great work by Bartłomiej Baranowski & Dave Zhenyu Chen

📢 OneCanvas: 3D Scene Understanding via Panoramic Reprojection We extract features from video frames and reproject them into one occlusion-free view of the whole scene that a 2D VLM reads just like a normal image. We can center this view on any viewpoint, including an agent's own pose for situated reasoning. The same projection lets us create spatial training tasks with no human annotation, solvable only by reasoning over the 3D positions of real object features placed on an otherwise empty canvas. The result is a stock 2D VLM that reasons in 3D, setting a new state of the art across spatial benchmarks at far less compute. 🌐 ▶️ Great work by Bartłomiej Baranowski & Dave Zhenyu Chen

Matthias Niessner

25,198 次观看 • 2 个月前

Female athletes are harmed by terrible NCAA policies that let men into women’s collegiate sports. There’s a man on San José State University volleyball team. This is a record of our dissent. 👀 2 young women grab the banner. Police intervene, say we had 1A rights. Later they approach & we talk

Female athletes are harmed by terrible NCAA policies that let men into women’s collegiate sports. There’s a man on San José State University volleyball team. This is a record of our dissent. 👀 2 young women grab the banner. Police intervene, say we had 1A rights. Later they approach & we talk

WomenAreReal

154,912 次观看 • 1 年前

We are currently meeting with the Governor of Shandong Province, Mr. Zhou Naixiang, as part of our efforts to foster a Shandong–Ogun State Economic and Trade Partnership under the broader Nigeria–China Strategic Partnership (NCSP). This engagement is a reflection of our unwavering commitment to expanding the frontiers of investment and cooperation for the benefit of our people. During our discussions, we are exploring opportunities in key sectors such as agriculture, manufacturing, technology, infrastructure, and human capital development. Shandong Province, being one of China’s most industrially advanced regions, presents immense prospects for collaboration, and we are confident that this partnership will open new doors for Ogun State in terms of trade, investment inflows, job creation, and skills transfer. As a government, we remain focused on positioning Ogun State as the gateway to Nigeria’s industrial future, and this meeting marks another milestone in our vision of building an economy that thrives on innovation, collaboration, and global partnerships. #BuildingOurFutureTogether #ISEYA #OgunState #DapoAbiodun #OgunChinaInvestmentDrive

We are currently meeting with the Governor of Shandong Province, Mr. Zhou Naixiang, as part of our efforts to foster a Shandong–Ogun State Economic and Trade Partnership under the broader Nigeria–China Strategic Partnership (NCSP). This engagement is a reflection of our unwavering commitment to expanding the frontiers of investment and cooperation for the benefit of our people. During our discussions, we are exploring opportunities in key sectors such as agriculture, manufacturing, technology, infrastructure, and human capital development. Shandong Province, being one of China’s most industrially advanced regions, presents immense prospects for collaboration, and we are confident that this partnership will open new doors for Ogun State in terms of trade, investment inflows, job creation, and skills transfer. As a government, we remain focused on positioning Ogun State as the gateway to Nigeria’s industrial future, and this meeting marks another milestone in our vision of building an economy that thrives on innovation, collaboration, and global partnerships. #BuildingOurFutureTogether #ISEYA #OgunState #DapoAbiodun #OgunChinaInvestmentDrive

Prince Dr. Dapo Abiodun, CON

19,195 次观看 • 10 个月前

Nicole Shanahan warns of an “extinction” level risk if AI-developed mRNA vaccines are rushed into use. “What we need for the mRNA platform right now is a moratorium. It's not ready for human use.” “It delivers an inconsistent result in individuals.” “In order for our population to grow, to be strong, to be fully able-bodied, and for our human economy to thrive, we do need a moratorium on the mRNA for the time being.” “AI is a computer system. Human health is not.” “There's so many that could have been shared in yesterday's conference that are really excellent uses of AI… I heard a few that were kind of out there and if deployed too quickly could lead to an extinction event.”

Nicole Shanahan warns of an “extinction” level risk if AI-developed mRNA vaccines are rushed into use. “What we need for the mRNA platform right now is a moratorium. It's not ready for human use.” “It delivers an inconsistent result in individuals.” “In order for our population to grow, to be strong, to be fully able-bodied, and for our human economy to thrive, we do need a moratorium on the mRNA for the time being.” “AI is a computer system. Human health is not.” “There's so many that could have been shared in yesterday's conference that are really excellent uses of AI… I heard a few that were kind of out there and if deployed too quickly could lead to an extinction event.”

End Tribalism in Politics

572,312 次观看 • 1 年前

TokenFlow: Consistent Diffusion Features for Consistent Video Editing paper page: The generative AI revolution has recently expanded to videos. Nevertheless, current state-of-the-art video models are still lagging behind image models in terms of visual quality and user control over the generated content. In this work, we present a framework that harnesses the power of a text-to-image diffusion model for the task of text-driven video editing. Specifically, given a source video and a target text-prompt, our method generates a high-quality video that adheres to the target text, while preserving the spatial layout and motion of the input video. Our method is based on a key observation that consistency in the edited video can be obtained by enforcing consistency in the diffusion feature space. We achieve this by explicitly propagating diffusion features based on inter-frame correspondences, readily available in the model. Thus, our framework does not require any training or fine-tuning, and can work in conjunction with any off-the-shelf text-to-image editing method. We demonstrate state-of-the-art editing results on a variety of real-world videos.

TokenFlow: Consistent Diffusion Features for Consistent Video Editing paper page: The generative AI revolution has recently expanded to videos. Nevertheless, current state-of-the-art video models are still lagging behind image models in terms of visual quality and user control over the generated content. In this work, we present a framework that harnesses the power of a text-to-image diffusion model for the task of text-driven video editing. Specifically, given a source video and a target text-prompt, our method generates a high-quality video that adheres to the target text, while preserving the spatial layout and motion of the input video. Our method is based on a key observation that consistency in the edited video can be obtained by enforcing consistency in the diffusion feature space. We achieve this by explicitly propagating diffusion features based on inter-frame correspondences, readily available in the model. Thus, our framework does not require any training or fine-tuning, and can work in conjunction with any off-the-shelf text-to-image editing method. We demonstrate state-of-the-art editing results on a variety of real-world videos.

AK

143,469 次观看 • 3 年前

Sam Altman: "We now have AI systems smart enough that our best scientists are saying this changes the game for science." "These models are discovering new knowledge on their own. This is the first time a tool humans have built is autonomously contributing to the sum total of human knowledge."

Sam Altman: "We now have AI systems smart enough that our best scientists are saying this changes the game for science." "These models are discovering new knowledge on their own. This is the first time a tool humans have built is autonomously contributing to the sum total of human knowledge."

vitrupo

26,992 次观看 • 3 个月前

“This is the first time in our lifetime that we are seeing a genocide live streamed to our phones every single day. You ask me, what does the global south & global north think of human rights? They have spat on human rights. They've spat on humanity” —Varsha Gandikota-Nellutla

“This is the first time in our lifetime that we are seeing a genocide live streamed to our phones every single day. You ask me, what does the global south & global north think of human rights? They have spat on human rights. They've spat on humanity” —Varsha Gandikota-Nellutla

The Resonance

96,983 次观看 • 1 年前

We’re thrilled to announce that 2 of our projects are nominated for E-Commerce Site of the Year on awwwards. and that our team is nominated for Studio of the Year for the second time in a row. Take a look at our reel to get a glimpse of the nominated projects & last year’s work!

We’re thrilled to announce that 2 of our projects are nominated for E-Commerce Site of the Year on awwwards. and that our team is nominated for Studio of the Year for the second time in a row. Take a look at our reel to get a glimpse of the nominated projects & last year’s work!

Antinomy Studio

18,451 次观看 • 3 年前