Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

a robot can recognize every object in a room and still not know how to open a single drawer knowing there's a cabinet is scene understanding. knowing which handle to grab, which direction to pull, and that it's a drawer not a door is affordance understanding almost no dataset... show more

harpreet

7,617 subscribers

18,702 views • 14 days ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

LLaVA-3D A Simple yet Effective Pathway to Empowering LMMs with 3D-awareness Recent advancements in Large Multimodal Models (LMMs) have greatly enhanced their proficiency in 2D visual understanding tasks, enabling them to effectively process and understand images and videos. However, the development of LMMs with 3D-awareness for 3D scene understanding has been hindered by the lack of large-scale 3D vision-language datasets and powerful 3D encoders. In this paper, we introduce a simple yet effective framework called LLaVA-3D. Leveraging the strong 2D understanding priors from LLaVA, our LLaVA-3D efficiently adapts LLaVA for 3D scene understanding without compromising 2D understanding capabilities. To achieve this, we employ a simple yet effective representation, 3D Patch, which connects 2D CLIP patch features with their corresponding positions in 3D space. By integrating the 3D Patches into 2D LMMs and employing joint 2D and 3D vision-language instruction tuning, we establish a unified architecture for both 2D image understanding and 3D scene understanding. Experimental results show that LLaVA-3D converges 3.5x faster than existing 3D LMMs when trained on 3D vision-language datasets. Moreover, LLaVA-3D not only achieves state-of-the-art performance across various 3D tasks but also maintains comparable 2D image understanding and vision-language conversation capabilities with LLaVA.

LLaVA-3D A Simple yet Effective Pathway to Empowering LMMs with 3D-awareness Recent advancements in Large Multimodal Models (LMMs) have greatly enhanced their proficiency in 2D visual understanding tasks, enabling them to effectively process and understand images and videos. However, the development of LMMs with 3D-awareness for 3D scene understanding has been hindered by the lack of large-scale 3D vision-language datasets and powerful 3D encoders. In this paper, we introduce a simple yet effective framework called LLaVA-3D. Leveraging the strong 2D understanding priors from LLaVA, our LLaVA-3D efficiently adapts LLaVA for 3D scene understanding without compromising 2D understanding capabilities. To achieve this, we employ a simple yet effective representation, 3D Patch, which connects 2D CLIP patch features with their corresponding positions in 3D space. By integrating the 3D Patches into 2D LMMs and employing joint 2D and 3D vision-language instruction tuning, we establish a unified architecture for both 2D image understanding and 3D scene understanding. Experimental results show that LLaVA-3D converges 3.5x faster than existing 3D LMMs when trained on 3D vision-language datasets. Moreover, LLaVA-3D not only achieves state-of-the-art performance across various 3D tasks but also maintains comparable 2D image understanding and vision-language conversation capabilities with LLaVA.

AK

41,729 views • 1 year ago

Scaling 3D scene data is a long-standing challenge in scene understanding, spatial reasoning, and robotics. Since scanning, reconstruction, and labeling are so labor-intensive, data scarcity has remained a major bottleneck. 🛑 To solve this, we propose SceneVerse++: Lifting Unlabeled Internet-level Data for 3D Scene Understanding (CVPR 2026). By reconstructing internet videos and annotating 3D scenes automatically, we’ve created a massive real-world dataset for end-to-end understanding. 🌐📐 SceneVerse++ makes it easy to scale "in-the-wild" 3D scenes toward more capable spatial reasoning systems. This significantly promotes progress in 3D VQA, visual navigation, and broader tasks in Embodied AI and Robotics. 🤖🦾 We are fully open-sourced! Check out the paper, code, and data here: 🌐 Project: 📄 Paper: 📊 Dataset: Code:

Scaling 3D scene data is a long-standing challenge in scene understanding, spatial reasoning, and robotics. Since scanning, reconstruction, and labeling are so labor-intensive, data scarcity has remained a major bottleneck. 🛑 To solve this, we propose SceneVerse++: Lifting Unlabeled Internet-level Data for 3D Scene Understanding (CVPR 2026). By reconstructing internet videos and annotating 3D scenes automatically, we’ve created a massive real-world dataset for end-to-end understanding. 🌐📐 SceneVerse++ makes it easy to scale "in-the-wild" 3D scenes toward more capable spatial reasoning systems. This significantly promotes progress in 3D VQA, visual navigation, and broader tasks in Embodied AI and Robotics. 🤖🦾 We are fully open-sourced! Check out the paper, code, and data here: 🌐 Project: 📄 Paper: 📊 Dataset: Code:

Siyuan Huang

12,612 views • 2 months ago

BREAKING: IRAN FM ARAGHCHI CONFIRMS 14 POINT MOU: "We set in motion a process that is now drawing close to its end. The result is a fourteen-point memorandum of understanding… it has not yet been signed, and until the very last moment things may still shift. Once it is done, I will come and explain every point to the people."

BREAKING: IRAN FM ARAGHCHI CONFIRMS 14 POINT MOU: "We set in motion a process that is now drawing close to its end. The result is a fourteen-point memorandum of understanding… it has not yet been signed, and until the very last moment things may still shift. Once it is done, I will come and explain every point to the people."

Sulaiman Ahmed

114,164 views • 1 month ago

Dr Fei-Fei-Li explains with a simple example how everyday household chores are so extremely difficult for Robots. "If you tell a robot to open the top drawer and watch out for the vase, this is actually a really hard task for robots." because the robot must ground language into the real world. Words like "top", "drawer", and "vase" are abstract. The system has to map them to 3D locations, objects, and relations in a noisy scene. This requires robust perception, object recognition, and spatial reasoning under uncertainty. The robot also lacks human commonsense. "Watch out" implies predicting consequences, estimating clearances, and understanding that vases are fragile. Encoding such priors, like how heavy a drawer is or how a vase might tip, is very complex and difficult without rich world knowledge. Learning the behavior from rewards is tough. The success signal is very sparse here, so naive exploration almost never stumbles on a full success sequence. This makes policy learning sample inefficient and brittle, especially when the environment changes between training and deployment. A sparse reward situation is when the agent only gets a success signal at the very end, and gets little or no feedback along the way. If a robot must open a drawer without hitting a vase, it might get reward only if the drawer ends up open and the vase is intact. Every partial try before that looks the same to the learner, reward equals 0. --- From "DSAI by Dr. Osbert Tay" YT channel

Dr Fei-Fei-Li explains with a simple example how everyday household chores are so extremely difficult for Robots. "If you tell a robot to open the top drawer and watch out for the vase, this is actually a really hard task for robots." because the robot must ground language into the real world. Words like "top", "drawer", and "vase" are abstract. The system has to map them to 3D locations, objects, and relations in a noisy scene. This requires robust perception, object recognition, and spatial reasoning under uncertainty. The robot also lacks human commonsense. "Watch out" implies predicting consequences, estimating clearances, and understanding that vases are fragile. Encoding such priors, like how heavy a drawer is or how a vase might tip, is very complex and difficult without rich world knowledge. Learning the behavior from rewards is tough. The success signal is very sparse here, so naive exploration almost never stumbles on a full success sequence. This makes policy learning sample inefficient and brittle, especially when the environment changes between training and deployment. A sparse reward situation is when the agent only gets a success signal at the very end, and gets little or no feedback along the way. If a robot must open a drawer without hitting a vase, it might get reward only if the drawer ends up open and the vase is intact. Every partial try before that looks the same to the learner, reward equals 0. --- From "DSAI by Dr. Osbert Tay" YT channel

Rohan Paul

342,553 views • 8 months ago

✨ Made a new mini feature on Photo AI: [ Grab from 3d model ] So the problem is we're at that stage in time (typical for AI) where image-to-3d models are not good enough but are fun to play with, but we know they'll be good enough in 1-2 years With [ Make 3d model ] you already can turn any Photo AI pic into a 3d model but it still looks hyper clunky and deformed, but it works! One cool idea I had to make that more useful and made now: Let people make a 3d model then change the view of the it with the 3d viewer, then press [ o ] and it grabs a frame of the 3d That image you can then [ Remix ] (img2img), and it becomes a real photo again and that in turn you can then turn into a video again with [ Make video ] So that essentially gives you a fully freeform camera position control to take photos with One thing I need to fix is the background/skybox, I kinda need to take the original photo and remove the person and just get the background for the 3d model viewer, in this case it should be white, but it's a start!

✨ Made a new mini feature on Photo AI: [ Grab from 3d model ] So the problem is we're at that stage in time (typical for AI) where image-to-3d models are not good enough but are fun to play with, but we know they'll be good enough in 1-2 years With [ Make 3d model ] you already can turn any Photo AI pic into a 3d model but it still looks hyper clunky and deformed, but it works! One cool idea I had to make that more useful and made now: Let people make a 3d model then change the view of the it with the 3d viewer, then press [ o ] and it grabs a frame of the 3d That image you can then [ Remix ] (img2img), and it becomes a real photo again and that in turn you can then turn into a video again with [ Make video ] So that essentially gives you a fully freeform camera position control to take photos with One thing I need to fix is the background/skybox, I kinda need to take the original photo and remove the person and just get the background for the 3d model viewer, in this case it should be white, but it's a start!

@levelsio

119,210 views • 1 year ago

THIS GUY VIBE CODED A TOOL THAT TURNS ANY SVG INTO A 3D OBJECT YOU CAN SPIN, ANIMATE, AND EMBED drag in an SVG, type some text, or draw pixel art. it becomes a 3D object instantly. spin it around, animate it, and embed it on your site. export as 4K image or video. runs entirely in your browser and nothing gets uploaded to any server with no account needed. 100% free AND its open source. this is one of those tools you didn't know you needed until you see it

THIS GUY VIBE CODED A TOOL THAT TURNS ANY SVG INTO A 3D OBJECT YOU CAN SPIN, ANIMATE, AND EMBED drag in an SVG, type some text, or draw pixel art. it becomes a 3D object instantly. spin it around, animate it, and embed it on your site. export as 4K image or video. runs entirely in your browser and nothing gets uploaded to any server with no account needed. 100% free AND its open source. this is one of those tools you didn't know you needed until you see it

Om Patel

229,955 views • 3 months ago

74 seconds to print a 3D Benchy! 🚢 Someone just printed a 3D Benchy in 74 seconds! For context: Benchy is the standard benchmark for 3D printing quality. Getting it done in 74 seconds required pushing every variable to its limit. Pulley and belt systems optimized for minimal slop, machine rigidity improvements to eliminate flex under high acceleration, input shaping to cancel out resonance frequencies that cause ringing artifacts at high speeds. A 3D scanner to analyze print quality objectively. Frequency analysis to identify and cancel specific resonance modes in the frame and motion system. Check it out here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

74 seconds to print a 3D Benchy! 🚢 Someone just printed a 3D Benchy in 74 seconds! For context: Benchy is the standard benchmark for 3D printing quality. Getting it done in 74 seconds required pushing every variable to its limit. Pulley and belt systems optimized for minimal slop, machine rigidity improvements to eliminate flex under high acceleration, input shaping to cancel out resonance frequencies that cause ringing artifacts at high speeds. A 3D scanner to analyze print quality objectively. Frequency analysis to identify and cancel specific resonance modes in the frame and motion system. Check it out here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

108,721 views • 2 months ago

We have seen a lot of legged robots doing navigation in the wild. But how about mobile manipulation in the wild? I have been pushing the direction of learning a unified, efficient, and dynamic 3D representation of scenes (for navigation) and objects (for manipulation) for the past two years. And now we have GeFF --- our large-scale, generalizable feature field, that combines the speed of a feed-forward neural network with the rich semantics from Foundation Models, to handle dynamically changing scenes, and enable open-ended, language-grounded scene and object understanding.

We have seen a lot of legged robots doing navigation in the wild. But how about mobile manipulation in the wild? I have been pushing the direction of learning a unified, efficient, and dynamic 3D representation of scenes (for navigation) and objects (for manipulation) for the past two years. And now we have GeFF --- our large-scale, generalizable feature field, that combines the speed of a feed-forward neural network with the rich semantics from Foundation Models, to handle dynamically changing scenes, and enable open-ended, language-grounded scene and object understanding.

Xiaolong Wang

42,767 views • 2 years ago

We shipped 3D AI Agents straight into Sperax | SperaxOS and they're alive inside the chat. Build your 3D AI agent on → animations, emotions, a brain, money. Paste the link → it renders live in the conversation. Not a video. A real, live, interactive 3D agent. Spin it. Emote it. Give it a wallet, skills, and memory. Put it to work. Time to break AI out the box.

We shipped 3D AI Agents straight into Sperax | SperaxOS and they're alive inside the chat. Build your 3D AI agent on → animations, emotions, a brain, money. Paste the link → it renders live in the conversation. Not a video. A real, live, interactive 3D agent. Spin it. Emote it. Give it a wallet, skills, and memory. Put it to work. Time to break AI out the box.

three.ws

52,085 views • 1 month ago

Apple Maps 3d is insanely impressive now (see video below) Google on the other hand decided to build an insanely clunky solution to display 3d maps to avoid using mobile GPUs: A Google server streams (!) a live video of its 3d rendered map to you, the result is a disastrous: a super laggy and clunky 3d view full of video compression artefacts with a response time of 2-3 seconds for every movement And I know WHY they built that, someone felt it was not inclusive to only have 3d maps that are high performing on iOS because Apple's GPUs are insanely fast and can deal with it (as you can see below in Apple Maps) but they wanted to support inferior phones, I get it But then they also decided to worsen the experience for iOS users who could easily run 3d natively on its GPU It's just an insane product decision, there's no way to cut it: they could have super smooth 3d like Apple Maps but neutered it with a clunky live stream Even more insane because Google does have native 3d on iOS in a completely different app that nobody uses: Google Earth Is 3d important? Not rly for nav no but it shows again something about Google's engineering decisions, like the redesign of the Google OAuth Login screen that made zero sense Nobody is speaking up against bad ideas in Google meetings, everyone's too scared of HR maybe, it's the opposite of a meritocracy and it's just super visible from using Google products right now

Apple Maps 3d is insanely impressive now (see video below) Google on the other hand decided to build an insanely clunky solution to display 3d maps to avoid using mobile GPUs: A Google server streams (!) a live video of its 3d rendered map to you, the result is a disastrous: a super laggy and clunky 3d view full of video compression artefacts with a response time of 2-3 seconds for every movement And I know WHY they built that, someone felt it was not inclusive to only have 3d maps that are high performing on iOS because Apple's GPUs are insanely fast and can deal with it (as you can see below in Apple Maps) but they wanted to support inferior phones, I get it But then they also decided to worsen the experience for iOS users who could easily run 3d natively on its GPU It's just an insane product decision, there's no way to cut it: they could have super smooth 3d like Apple Maps but neutered it with a clunky live stream Even more insane because Google does have native 3d on iOS in a completely different app that nobody uses: Google Earth Is 3d important? Not rly for nav no but it shows again something about Google's engineering decisions, like the redesign of the Google OAuth Login screen that made zero sense Nobody is speaking up against bad ideas in Google meetings, everyone's too scared of HR maybe, it's the opposite of a meritocracy and it's just super visible from using Google products right now

@levelsio

1,488,855 views • 1 year ago

Sora's videos look impressive, but physics understanding and consistancy is still not there yet. It was a quite simple script: "A humanoid robot standing near the table with red, green and blue cubes on it performs a cube stacking task, with red in the bottom and blue on the top." Pretty impressive that it still managed to achieve the goal, only not in a very usual way.

Sora's videos look impressive, but physics understanding and consistancy is still not there yet. It was a quite simple script: "A humanoid robot standing near the table with red, green and blue cubes on it performs a cube stacking task, with red in the bottom and blue on the top." Pretty impressive that it still managed to achieve the goal, only not in a very usual way.

ViktorM🇺🇦

51,608 views • 1 year ago

New open-source 3D world-generation model. I'm rendering a couple of worlds in the video, so check it out. You'll find the GitHub and the Hugging Face links to the model below. This is a multi-modal world model that you can use for a bunch of things: • To generate new worlds • To reconstruct worlds • To simulate 3D interactive worlds from a prompt, images, or a video You can edit the 3D outputs in Unity and Unreal Engine (they export as meshes, 3DGS files, and point clouds). You can also generate 3D characters in the world and walk around. Pretty fun stuff!

New open-source 3D world-generation model. I'm rendering a couple of worlds in the video, so check it out. You'll find the GitHub and the Hugging Face links to the model below. This is a multi-modal world model that you can use for a bunch of things: • To generate new worlds • To reconstruct worlds • To simulate 3D interactive worlds from a prompt, images, or a video You can edit the 3D outputs in Unity and Unreal Engine (they export as meshes, 3DGS files, and point clouds). You can also generate 3D characters in the world and walk around. Pretty fun stuff!

Santiago

65,446 views • 2 months ago

📢 Lyra: Generative 3D Scene Reconstruction via Video Diffusion Model Self-Distillation Got only one or a few images and wondering if recovering the 3D environment is a reconstruction or generation problem? Why not do it with a generative reconstruction model! We show that a camera-conditioned video diffusion model can be transformed into a generative reconstruction model that directly outputs a high-quality 3D Gaussian Splatting representation through self-distillation, without requiring real-world training data. Check out our results in the video (wait for dynamic scenes in the second half!) : Project Page: Code and Models: Paper:

📢 Lyra: Generative 3D Scene Reconstruction via Video Diffusion Model Self-Distillation Got only one or a few images and wondering if recovering the 3D environment is a reconstruction or generation problem? Why not do it with a generative reconstruction model! We show that a camera-conditioned video diffusion model can be transformed into a generative reconstruction model that directly outputs a high-quality 3D Gaussian Splatting representation through self-distillation, without requiring real-world training data. Check out our results in the video (wait for dynamic scenes in the second half!) : Project Page: Code and Models: Paper:

Sherwin Bahmani

66,563 views • 9 months ago

WOW. 😳 Apple just quietly won the 3D maps war at WWDC. Gaussian Splatting is coming to Apple Maps Flyover this fall. Apple Maps Flyover covers 300+ cities. Until yesterday, every single one was built on standard drone photogrammetry. The technology captures photos from the air and reconstructs 3D geometry from them. Gaussian Splatting does not reconstruct geometry. It represents the scene as millions of tiny 3D ellipsoids, each one carrying its own color and opacity information based on how light actually behaves in that location. The output is not a mesh model. It is a field of light. When you move through it, it does not crumble at the edges. The detail holds because it was never geometry to begin with. Apple has been hiring for this for years. Their SHARP model, published in research last year, generates photorealistic 3D scenes from a single image in under a second. Google has more sensor data than anyone. More Street View cars, more satellites, more capture history. On navigation accuracy and geodata depth, Google Maps is still ahead by most measures. But fidelity in 3D city rendering is a different competition, and Apple just set a bar in that. Most people will experience this in the fall without knowing the name of the technology. They will open Flyover, look at a city they know, and notice it looks different. Real, not rendered. That is the moment Gaussian Splatting stops being a research term and becomes something a billion people use. Bookmark this. It will look prescient by October.

WOW. 😳 Apple just quietly won the 3D maps war at WWDC. Gaussian Splatting is coming to Apple Maps Flyover this fall. Apple Maps Flyover covers 300+ cities. Until yesterday, every single one was built on standard drone photogrammetry. The technology captures photos from the air and reconstructs 3D geometry from them. Gaussian Splatting does not reconstruct geometry. It represents the scene as millions of tiny 3D ellipsoids, each one carrying its own color and opacity information based on how light actually behaves in that location. The output is not a mesh model. It is a field of light. When you move through it, it does not crumble at the edges. The detail holds because it was never geometry to begin with. Apple has been hiring for this for years. Their SHARP model, published in research last year, generates photorealistic 3D scenes from a single image in under a second. Google has more sensor data than anyone. More Street View cars, more satellites, more capture history. On navigation accuracy and geodata depth, Google Maps is still ahead by most measures. But fidelity in 3D city rendering is a different competition, and Apple just set a bar in that. Most people will experience this in the fall without knowing the name of the technology. They will open Flyover, look at a city they know, and notice it looks different. Real, not rendered. That is the moment Gaussian Splatting stops being a research term and becomes something a billion people use. Bookmark this. It will look prescient by October.

Shruti

19,784 views • 1 month ago

More and more users are turning their 360 footage into fully explorable 3D Gaussian Splats with View it here: This reconstruction started from a simple capture and is now a fully explorable 3D Gaussian Splat rendered in the FreeGaussian viewer, complete with point cloud reconstruction and high-fidelity spatial detail. More and more creators are discovering that their existing footage can become immersive 3D scenes. And they’re doing it for free with FreeGaussian

More and more users are turning their 360 footage into fully explorable 3D Gaussian Splats with View it here: This reconstruction started from a simple capture and is now a fully explorable 3D Gaussian Splat rendered in the FreeGaussian viewer, complete with point cloud reconstruction and high-fidelity spatial detail. More and more creators are discovering that their existing footage can become immersive 3D scenes. And they’re doing it for free with FreeGaussian

Over the Reality 🌐

14,682 views • 2 months ago

Yes, AI is here and 3D artists have to adapt. Here's one way I'm integrating AI with my 3D work, as a set extension. The BG is generated by Kling 3.0 and brought into my 3D scene so I can fully control the timing of the lights in the FG, which is 3D. Seedance 2.0 next!!

Yes, AI is here and 3D artists have to adapt. Here's one way I'm integrating AI with my 3D work, as a set extension. The BG is generated by Kling 3.0 and brought into my 3D scene so I can fully control the timing of the lights in the FG, which is 3D. Seedance 2.0 next!!

David Ariew

29,651 views • 4 months ago

Blended-NeRF: Zero-Shot Object Generation and Blending in Existing Neural Radiance Fields paper page: Editing a local region or a specific object in a 3D scene represented by a NeRF is challenging, mainly due to the implicit nature of the scene representation. Consistently blending a new realistic object into the scene adds an additional level of difficulty. We present Blended-NeRF, a robust and flexible framework for editing a specific region of interest in an existing NeRF scene, based on text prompts or image patches, along with a 3D ROI box. Our method leverages a pretrained language-image model to steer the synthesis towards a user-provided text prompt or image patch, along with a 3D MLP model initialized on an existing NeRF scene to generate the object and blend it into a specified region in the original scene. We allow local editing by localizing a 3D ROI box in the input scene, and seamlessly blend the content synthesized inside the ROI with the existing scene using a novel volumetric blending technique. To obtain natural looking and view-consistent results, we leverage existing and new geometric priors and 3D augmentations for improving the visual fidelity of the final result. We test our framework both qualitatively and quantitatively on a variety of real 3D scenes and text prompts, demonstrating realistic multi-view consistent results with much flexibility and diversity compared to the baselines. Finally, we show the applicability of our framework for several 3D editing applications, including adding new objects to a scene, removing/replacing/altering existing objects, and texture conversion.

Blended-NeRF: Zero-Shot Object Generation and Blending in Existing Neural Radiance Fields paper page: Editing a local region or a specific object in a 3D scene represented by a NeRF is challenging, mainly due to the implicit nature of the scene representation. Consistently blending a new realistic object into the scene adds an additional level of difficulty. We present Blended-NeRF, a robust and flexible framework for editing a specific region of interest in an existing NeRF scene, based on text prompts or image patches, along with a 3D ROI box. Our method leverages a pretrained language-image model to steer the synthesis towards a user-provided text prompt or image patch, along with a 3D MLP model initialized on an existing NeRF scene to generate the object and blend it into a specified region in the original scene. We allow local editing by localizing a 3D ROI box in the input scene, and seamlessly blend the content synthesized inside the ROI with the existing scene using a novel volumetric blending technique. To obtain natural looking and view-consistent results, we leverage existing and new geometric priors and 3D augmentations for improving the visual fidelity of the final result. We test our framework both qualitatively and quantitatively on a variety of real 3D scenes and text prompts, demonstrating realistic multi-view consistent results with much flexibility and diversity compared to the baselines. Finally, we show the applicability of our framework for several 3D editing applications, including adding new objects to a scene, removing/replacing/altering existing objects, and texture conversion.

AK

62,768 views • 3 years ago

PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Tracking paper page: introduce PointOdyssey, a large-scale synthetic dataset, and data generation framework, for the training and evaluation of long-term fine-grained tracking algorithms. Our goal is to advance the state-of-the-art by placing emphasis on long videos with naturalistic motion. Toward the goal of naturalism, we animate deformable characters using real-world motion capture data, we build 3D scenes to match the motion capture environments, and we render camera viewpoints using trajectories mined via structure-from-motion on real videos. We create combinatorial diversity by randomizing character appearance, motion profiles, materials, lighting, 3D assets, and atmospheric effects. Our dataset currently includes 104 videos, averaging 2,000 frames long, with orders of magnitude more correspondence annotations than prior work. We show that existing methods can be trained from scratch in our dataset and outperform the published variants. Finally, we introduce modifications to the PIPs point tracking method, greatly widening its temporal receptive field, which improves its performance on PointOdyssey as well as on two real-world benchmarks.

PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Tracking paper page: introduce PointOdyssey, a large-scale synthetic dataset, and data generation framework, for the training and evaluation of long-term fine-grained tracking algorithms. Our goal is to advance the state-of-the-art by placing emphasis on long videos with naturalistic motion. Toward the goal of naturalism, we animate deformable characters using real-world motion capture data, we build 3D scenes to match the motion capture environments, and we render camera viewpoints using trajectories mined via structure-from-motion on real videos. We create combinatorial diversity by randomizing character appearance, motion profiles, materials, lighting, 3D assets, and atmospheric effects. Our dataset currently includes 104 videos, averaging 2,000 frames long, with orders of magnitude more correspondence annotations than prior work. We show that existing methods can be trained from scratch in our dataset and outperform the published variants. Finally, we introduce modifications to the PIPs point tracking method, greatly widening its temporal receptive field, which improves its performance on PointOdyssey as well as on two real-world benchmarks.

AK

122,533 views • 2 years ago

I took Clicky and I made it 5x Faster So I saw that Farza 🇵🇰🇺🇸 was using uses Claude's vision to find UI elements on screen, send a screenshot, wait for coordinates back. It works, but it's slow. I replaced that with OmniParser V2 by Microsoft which is a local Object detection model trained specifically on UI elements. It runs on-device, detects every button, menu, and icon in 400ms, and gives me pixel-precise coordinates. No API call, no latency, no cost. The green highlights you see around the UI elements is the detection overlay and you can see as I am switching the screen it takes no time to detect and highlight which is pretty neat! With Local models improving by the day, Its the right direction for applications like these. Next up: video-synced tutorials where a YouTube tutorial pauses and waits for you to perform each action in the real app. I am not stopping!

I took Clicky and I made it 5x Faster So I saw that Farza 🇵🇰🇺🇸 was using uses Claude's vision to find UI elements on screen, send a screenshot, wait for coordinates back. It works, but it's slow. I replaced that with OmniParser V2 by Microsoft which is a local Object detection model trained specifically on UI elements. It runs on-device, detects every button, menu, and icon in 400ms, and gives me pixel-precise coordinates. No API call, no latency, no cost. The green highlights you see around the UI elements is the detection overlay and you can see as I am switching the screen it takes no time to detect and highlight which is pretty neat! With Local models improving by the day, Its the right direction for applications like these. Next up: video-synced tutorials where a YouTube tutorial pauses and waits for you to perform each action in the real app. I am not stopping!

Milind S

73,026 views • 3 months ago