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。

Tomorrow I'll talk at the #ICRA workshop on "Enhancing human mobility" on recent 3D human estimation works from our group led by Gerard Pons-Moll. And I've released PyTorch training code for NLF, and more tools. Here's a preview of the even stronger upcoming DINOv2-ViTg-based NLF

István Sárándi

1,178 subscribers

12,124 次观看 • 1 年前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

I'll present #NLF, our SOTA 3D human pose and shape estimator at #NeurIPS2024 today: poster at East Exhibit Hall A-C #1304 from 11:00! Project: Models out for both PyTorch and TF! Further details to come soon!

I'll present #NLF, our SOTA 3D human pose and shape estimator at #NeurIPS2024 today: poster at East Exhibit Hall A-C #1304 from 11:00! Project: Models out for both PyTorch and TF! Further details to come soon!

István Sárándi

79,755 次观看 • 1 年前

WHAM defines the new state of the art in 3D human pose estimation from video. By a large margin. It’s fast, accurate, and it computes human pose in world coordinates. It’s also the first video-based method to be more accurate than single-image methods. 1/8

WHAM defines the new state of the art in 3D human pose estimation from video. By a large margin. It’s fast, accurate, and it computes human pose in world coordinates. It’s also the first video-based method to be more accurate than single-image methods. 1/8

Michael Black

118,463 次观看 • 2 年前

The BEDLAM2.0 dataset (B2) is here, just in time to train your 3D human pose and shape estimation methods for CVPR. B2 goes beyond BEDLAM (B1) to include widely varied and natural camera motions and fields of view, more diverse body shapes, strand-based hair, more garments, shoes, more body motions, and more 3D scenes. Compared with B1, training on B2 produces more accurate 3D human pose, resulting in SOTA accuracy, particularly for estimates in world coordinates. B2 lets you jointly train camera motion and human motion regressors, and we also provide depth maps. Check out data, code, dataset statistics, and much more. BEDLAM2.0 will appear in the 2025 NeurIPS Datasets and Benchmarks Track. Joint work with Joachim Tesch, Giorgio Becherini, Prerana Achar, Anastasios Yiannakidis, Muhammed Kocabas, Priyanka Patel.

The BEDLAM2.0 dataset (B2) is here, just in time to train your 3D human pose and shape estimation methods for CVPR. B2 goes beyond BEDLAM (B1) to include widely varied and natural camera motions and fields of view, more diverse body shapes, strand-based hair, more garments, shoes, more body motions, and more 3D scenes. Compared with B1, training on B2 produces more accurate 3D human pose, resulting in SOTA accuracy, particularly for estimates in world coordinates. B2 lets you jointly train camera motion and human motion regressors, and we also provide depth maps. Check out data, code, dataset statistics, and much more. BEDLAM2.0 will appear in the 2025 NeurIPS Datasets and Benchmarks Track. Joint work with Joachim Tesch, Giorgio Becherini, Prerana Achar, Anastasios Yiannakidis, Muhammed Kocabas, Priyanka Patel.

Michael Black

27,523 次观看 • 8 个月前

Happy to share the publication of our paper in nature Conformational ensembles of the human intrinsically disordered proteome Work led by Giulio Tesei and Anna Ida🕊. I'll post more later, but for now here is a link: and a short movie about the work

Happy to share the publication of our paper in nature Conformational ensembles of the human intrinsically disordered proteome Work led by Giulio Tesei and Anna Ida🕊. I'll post more later, but for now here is a link: and a short movie about the work

Kresten Lindorff-Larsen

119,691 次观看 • 2 年前

My first attempt at putting a talk up directly here (I usually just add links to the video on our channel Here goes: this was a ~55 minute presentation and I had been asked to talk about the "Multiscale Human" (for I cover the topics of "what are we", "implications for biomedicine", and "future beyond human repair", from the perspectives of diverse intelligence at different spatial scales of our bodies and at the different time scales of evolution, development, and cognition.

My first attempt at putting a talk up directly here (I usually just add links to the video on our channel Here goes: this was a ~55 minute presentation and I had been asked to talk about the "Multiscale Human" (for I cover the topics of "what are we", "implications for biomedicine", and "future beyond human repair", from the perspectives of diverse intelligence at different spatial scales of our bodies and at the different time scales of evolution, development, and cognition.

Michael Levin

55,275 次观看 • 1 年前

We have received many recent media enquiries asking how a #heatwave impacts the human body. Here is a short animation from Heat and Health Research Incubator that explains the physiological impacts based on our 2021 paper The Lancet: University of Sydney #LeadershipForGood

We have received many recent media enquiries asking how a #heatwave impacts the human body. Here is a short animation from Heat and Health Research Incubator that explains the physiological impacts based on our 2021 paper The Lancet: University of Sydney #LeadershipForGood

Ollie Jay

14,484 次观看 • 3 年前

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

151,511 次观看 • 1 年前

.Reddit, Inc. CCO: Tech is rising — trust in humans isn’t going anywhere. Based on recent polling: 73% say technology is getting in the way of human connection, and 65% people believe advice more when they get it from a real human. “Humanity is indelible.” #AxiosLive

.Reddit, Inc. CCO: Tech is rising — trust in humans isn’t going anywhere. Based on recent polling: 73% say technology is getting in the way of human connection, and 65% people believe advice more when they get it from a real human. “Humanity is indelible.” #AxiosLive

Axios

13,232 次观看 • 2 个月前

Sorry Eric and Andrew, the Bronx is ready for socialism! Tomorrow at 3 pm, I'll be at 861 Gerard Ave in the Bronx to talk about how we're taking our stolen wealth back from the billionaire bosses and landlords. Huge shoutout to Julie and the rest of our campaign team for helping us build a movement for housing, fair wages, and a future that belongs to us.

Sorry Eric and Andrew, the Bronx is ready for socialism! Tomorrow at 3 pm, I'll be at 861 Gerard Ave in the Bronx to talk about how we're taking our stolen wealth back from the billionaire bosses and landlords. Huge shoutout to Julie and the rest of our campaign team for helping us build a movement for housing, fair wages, and a future that belongs to us.

Andre Easton

55,065 次观看 • 8 个月前

Multi-view Reconstruction via SfM-guided Monocular Depth Estimation Contributions: • We propose a novel approach to inject SfM priors into diffusion-based depth estimation, enabling highly accurate and multi-view consistent depth predictions for each viewpoint. • Based on the proposed depth estimator, we design a new multi-view 3D geometry reconstruction framework and process some synthetic datasets to facilitate training. • We evaluate our method on diverse real-world scene data, including objects, indoor environments, streetscapes, and aerial scenes, demonstrating the superior performance and generalization capability of our approach.

Multi-view Reconstruction via SfM-guided Monocular Depth Estimation Contributions: • We propose a novel approach to inject SfM priors into diffusion-based depth estimation, enabling highly accurate and multi-view consistent depth predictions for each viewpoint. • Based on the proposed depth estimator, we design a new multi-view 3D geometry reconstruction framework and process some synthetic datasets to facilitate training. • We evaluate our method on diverse real-world scene data, including objects, indoor environments, streetscapes, and aerial scenes, demonstrating the superior performance and generalization capability of our approach.

MrNeRF

25,651 次观看 • 1 年前

“What’s happening inside Suwayda right now is darker, more vile, and more revolting than anything we’ve seen in the past seven days.” A thread based on direct testimony from the head of the Syrian Observatory for Human Rights. #Suwayda #Syria

“What’s happening inside Suwayda right now is darker, more vile, and more revolting than anything we’ve seen in the past seven days.” A thread based on direct testimony from the head of the Syrian Observatory for Human Rights. #Suwayda #Syria

Karim Franceschi

16,863 次观看 • 1 年前

Egocentric Videos for Robot Training 🧵 Researchers at NYU and UC Berkeley published research where they developed a system called EgoZero that trains robots using human demonstration videos recorded with smart glasses. 👓 The system converts first-person human actions into 3D point-based state-action representations. The policies executed on a gripper-equipped robot, achieved a 70% zero-shot success rate across seven manipulation tasks, with only 20 minutes of human data per task. EgoZero stands as one of the solid empirical proofs that egocentric smart-glass video collected from everyday human behavior can serve as powerful, scalable training data for real robot learning. Vincent Liu Ademi Adeniji

Egocentric Videos for Robot Training 🧵 Researchers at NYU and UC Berkeley published research where they developed a system called EgoZero that trains robots using human demonstration videos recorded with smart glasses. 👓 The system converts first-person human actions into 3D point-based state-action representations. The policies executed on a gripper-equipped robot, achieved a 70% zero-shot success rate across seven manipulation tasks, with only 20 minutes of human data per task. EgoZero stands as one of the solid empirical proofs that egocentric smart-glass video collected from everyday human behavior can serve as powerful, scalable training data for real robot learning. Vincent Liu Ademi Adeniji

VaderResearch

23,055 次观看 • 10 个月前

When people talk about AI, they often focus on the models. But behind every AI system are millions of human contributions. The data comes from human writing, speech, images, and behavior. The labels that help models learn patterns are created by people. The feedback that improves responses comes from people. And the moderation that keeps systems safe is handled by people. AI may look automated on the surface, but human intelligence powers every layer underneath it. Really enjoyed hearing Ishank Gupta and Maria Vovchok discuss this reality and KGeN's role in the evolving AI ecosystem during the recent X Space hosted by Gate. A great reminder that the future of AI is still deeply connected to human participation.

When people talk about AI, they often focus on the models. But behind every AI system are millions of human contributions. The data comes from human writing, speech, images, and behavior. The labels that help models learn patterns are created by people. The feedback that improves responses comes from people. And the moderation that keeps systems safe is handled by people. AI may look automated on the surface, but human intelligence powers every layer underneath it. Really enjoyed hearing Ishank Gupta and Maria Vovchok discuss this reality and KGeN's role in the evolving AI ecosystem during the recent X Space hosted by Gate. A great reminder that the future of AI is still deeply connected to human participation.

Bahdfella™ ❤️‍🔥

36,062 次观看 • 24 天前

‘For a Government that is full of human rights lawyers, why do the opinions and human rights of indigenous Chagossians not matter at all?’ Reform UK Nigel Farage grills Keir Starmer on the removal order of Chagossians from the Chagos Islands by the Labour Government.

‘For a Government that is full of human rights lawyers, why do the opinions and human rights of indigenous Chagossians not matter at all?’ Reform UK Nigel Farage grills Keir Starmer on the removal order of Chagossians from the Chagos Islands by the Labour Government.

GB News

32,416 次观看 • 4 个月前

New short course on Reinforcement Learning from Human Feedback! RLHF is one of the key techniques that led to the rise of modern LLMs. It is used to align LLMs with human preferences, to make them more honest, helpful and harmless, by (i) learning a reward function that mimics human preferences, as expressed in human-provided labels, then, (ii) tuning an LLM to generate outputs that receive a high reward. In this course, taught by Nikita Namjoshi, Developer Advocate for GenAI at Google Cloud, you'll learn the details of how RLHF works, including how to apply it to tune an LLM for your own applications. You'll also use an open source library to tune a base LLM to align with human preferences expressed in a training set, and evaluate the tuned model by comparing its responses before and after RLHF-tuning. Please sign up here!

New short course on Reinforcement Learning from Human Feedback! RLHF is one of the key techniques that led to the rise of modern LLMs. It is used to align LLMs with human preferences, to make them more honest, helpful and harmless, by (i) learning a reward function that mimics human preferences, as expressed in human-provided labels, then, (ii) tuning an LLM to generate outputs that receive a high reward. In this course, taught by Nikita Namjoshi, Developer Advocate for GenAI at Google Cloud, you'll learn the details of how RLHF works, including how to apply it to tune an LLM for your own applications. You'll also use an open source library to tune a base LLM to align with human preferences expressed in a training set, and evaluate the tuned model by comparing its responses before and after RLHF-tuning. Please sign up here!

Andrew Ng

205,537 次观看 • 2 年前

📢Happy to present Convex Splatting, a novel way for 3D reconstruction based on 3D smooth convexes. For the first time, a splatting-based method reaches the quality of NeRF sota methods but with real-time rendering and few primitives!! I expect this to replace Gaussian splatting for 3D in the coming months. CODE RELEASED TODAY! joint work with collaborators from Université de Liège Visual Geometry Group (VGG) , KAUST Computer Vision Lab (IVUL) a thread 🧵 1/n

📢Happy to present Convex Splatting, a novel way for 3D reconstruction based on 3D smooth convexes. For the first time, a splatting-based method reaches the quality of NeRF sota methods but with real-time rendering and few primitives!! I expect this to replace Gaussian splatting for 3D in the coming months. CODE RELEASED TODAY! joint work with collaborators from Université de Liège Visual Geometry Group (VGG) , KAUST Computer Vision Lab (IVUL) a thread 🧵 1/n

Abdullah Hamdi

71,661 次观看 • 1 年前

What happens when robot world models learn from human experience at scale? 🤔 DreamDojo from NVIDIA Research is a generalist robot world model pretrained on 44K hours of egocentric human videos and then post-trained on robot data to generalize across new objects and environments. After distillation, it runs at 10 FPS for live teleoperation, policy evaluation, and model-based planning. Read the ICML paper to learn more 📄

What happens when robot world models learn from human experience at scale? 🤔 DreamDojo from NVIDIA Research is a generalist robot world model pretrained on 44K hours of egocentric human videos and then post-trained on robot data to generalize across new objects and environments. After distillation, it runs at 10 FPS for live teleoperation, policy evaluation, and model-based planning. Read the ICML paper to learn more 📄

NVIDIA Robotics

21,506 次观看 • 12 天前

I made new pants for our 3D customizable vtuber model's next update! -^u^- Also, here are all of the hair options that have been released so far <3 Next upcoming update will include even more and expand on this selection.

I made new pants for our 3D customizable vtuber model's next update! -^u^- Also, here are all of the hair options that have been released so far <3 Next upcoming update will include even more and expand on this selection.

Pastell & Palette🤍 3D Customizable Model!

13,594 次观看 • 5 个月前

Dieter ~ A Panoply Curiosity Curiosities is an on-going series that explores a variety of ideas and concepts from the artistic to the technical. Audio by Human Robot Soul For more in the series -

Dieter ~ A Panoply Curiosity Curiosities is an on-going series that explores a variety of ideas and concepts from the artistic to the technical. Audio by Human Robot Soul For more in the series -

Panoply

14,109 次观看 • 2 年前