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.

Real-time object detection will never be the same. Traditional YOLO needs NMS to remove duplicate boxes; it's slow and inconsistent. YOLO26 skips it entirely: single-pass predictions, faster inference and up to 300 detections per image. Download model:

Akshay 🚀

280,813 subscribers

374,108 views • 4 months ago •via X (Twitter)

Education Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Detect worker falls in real time with Ultralytics YOLO26! 📦 Combine object detection and pose estimation to identify when a person carrying luggage boxes falls, helping logistics teams improve warehouse safety and respond faster to incidents. Get started ➡️ #Ultralytics #YOLO26 #AI #Logistics #WorkplaceSafety

Detect worker falls in real time with Ultralytics YOLO26! 📦 Combine object detection and pose estimation to identify when a person carrying luggage boxes falls, helping logistics teams improve warehouse safety and respond faster to incidents. Get started ➡️ #Ultralytics #YOLO26 #AI #Logistics #WorkplaceSafety

Ultralytics

16,029 views • 2 months ago

Pothole detection on the road in real time using Ultralytics YOLO26! 🕳️ Manual road inspections are slow, costly, and hard to scale. With object detection, potholes can be identified directly from street-level images or video feeds, enabling faster and more consistent road condition monitoring. How I built this demo: ✅ Trained a segmentation model on a custom dataset. ✅ Generated mask contours for each pothole. ✅ Leveraged the onnx-exported model for faster processing. #Pothole #RoadDamage #AI

Pothole detection on the road in real time using Ultralytics YOLO26! 🕳️ Manual road inspections are slow, costly, and hard to scale. With object detection, potholes can be identified directly from street-level images or video feeds, enabling faster and more consistent road condition monitoring. How I built this demo: ✅ Trained a segmentation model on a custom dataset. ✅ Generated mask contours for each pothole. ✅ Leveraged the onnx-exported model for faster processing. #Pothole #RoadDamage #AI

Muhammad Rizwan Munawar

30,677 views • 4 months ago

Meta's SAM 3 is a beast. before: collect data, train custom object detector, use tracker to estimate object motion - days now: track anything with text prompt - seconds real-time object detection will never be the same.

Meta's SAM 3 is a beast. before: collect data, train custom object detector, use tracker to estimate object motion - days now: track anything with text prompt - seconds real-time object detection will never be the same.

Akshay 🚀

31,412 views • 8 months ago

Football players detection + tracking using Ultralytics YOLO26 ⚽️ It took me 30 minutes to train the YOLO26 model on a custom dataset using Ultralytics Platform. Later, I used the object tracker (Bytetrack) for tracking the players; tracking could be improved further, and I am working on it this week. More info 👇 #soccer #football #MachineLearning

Football players detection + tracking using Ultralytics YOLO26 ⚽️ It took me 30 minutes to train the YOLO26 model on a custom dataset using Ultralytics Platform. Later, I used the object tracker (Bytetrack) for tracking the players; tracking could be improved further, and I am working on it this week. More info 👇 #soccer #football #MachineLearning

Muhammad Rizwan Munawar

42,504 views • 2 months ago

Train a TensorFlow object detection model – then deploy it on a robot 🤖 Iulia Feroli (Iulia Feroli) shows how to turn a notebook into a real-time object detection app. This tutorial works for any project – though we demonstrate the deployment on (and assisted by!) the #ReachyMini, an open-source robot from Pollen Robotics. Built with PyCharm + Claude Code. 👉 Watch it in action:

Train a TensorFlow object detection model – then deploy it on a robot 🤖 Iulia Feroli (Iulia Feroli) shows how to turn a notebook into a real-time object detection app. This tutorial works for any project – though we demonstrate the deployment on (and assisted by!) the #ReachyMini, an open-source robot from Pollen Robotics. Built with PyCharm + Claude Code. 👉 Watch it in action:

PyCharm, a JetBrains IDE

34,908 views • 2 months ago

PAYING PER MODEL IS THE DUMBEST THING IN TECH RIGHT NOW i was paying 3x what i needed to for AI inference the grid lets you buy a quality spec instead of a specific model.. it routes every request in real time to the cheapest option that qualifies swap one url and your code keeps working exactly the same openai-compatible, one line to switch, 200M free tokens to start

PAYING PER MODEL IS THE DUMBEST THING IN TECH RIGHT NOW i was paying 3x what i needed to for AI inference the grid lets you buy a quality spec instead of a specific model.. it routes every request in real time to the cheapest option that qualifies swap one url and your code keeps working exactly the same openai-compatible, one line to switch, 200M free tokens to start

Robin Delta

15,729 views • 2 months ago

NOT SURE HOW DUBAI WILL GLOSS OVER THESE IMAGES AND VIDEOS… It can never be the same old business as usual cosying up to Israel and America. THE ABRAHAM ACCORDS WILL NEED TO BE BURIED SIX FEET UNDER‼️

NOT SURE HOW DUBAI WILL GLOSS OVER THESE IMAGES AND VIDEOS… It can never be the same old business as usual cosying up to Israel and America. THE ABRAHAM ACCORDS WILL NEED TO BE BURIED SIX FEET UNDER‼️

Earth Hippy 🌎🕊️💚

34,425 views • 4 months ago

Holy sh!t ! OpenAI will have their custom inference chips ready in just a few months and deployed at scale by the end of the year! 🤯 Training chip = The heavy lifters that require massive amounts of data and power to build and teach the AI models from scratch. Inference chip = The specialized, highly efficient chips that actually run the AI and generate the answers in real-time when you use it. This is going to help OpenAI drastically cut down their massive compute costs, speed up model reasoning times, and finally break free from relying entirely on Nvidia to scale their operations.

Holy sh!t ! OpenAI will have their custom inference chips ready in just a few months and deployed at scale by the end of the year! 🤯 Training chip = The heavy lifters that require massive amounts of data and power to build and teach the AI models from scratch. Inference chip = The specialized, highly efficient chips that actually run the AI and generate the answers in real-time when you use it. This is going to help OpenAI drastically cut down their massive compute costs, speed up model reasoning times, and finally break free from relying entirely on Nvidia to scale their operations.

Chris

60,278 views • 4 months ago

this is the worst local AI will ever be. tomorrow it gets faster. next month the models get smarter. next year your GPU runs what a data center runs today. Qwen3.5-35B-A3B on a single 3090. told it to visualize its own expert routing. 256 experts, 8 active per token, rendered in 3D on the same GPU running inference. no API key. no subscription. no permission needed. closed AI isn't losing ground. it's losing the argument.

this is the worst local AI will ever be. tomorrow it gets faster. next month the models get smarter. next year your GPU runs what a data center runs today. Qwen3.5-35B-A3B on a single 3090. told it to visualize its own expert routing. 256 experts, 8 active per token, rendered in 3D on the same GPU running inference. no API key. no subscription. no permission needed. closed AI isn't losing ground. it's losing the argument.

Sudo su

106,710 views • 5 months ago

✨ From Mesh to 3D Gaussians - Shaded in Deferred Style ✨ By Stefano Scolari (with link to repo in the comment) When converting a mesh to 3DGS model, the Gaussians are embedded with PBR properties and normal info during conversion. This prepares them for shading, similar to traditional geometry. In this scenario, Stefano used a deferred approach for shading to demonstrate how seamlessly it translates from triangle-based pipelines. 🔥 Fun fact: The mesh-to-splat conversion only needs to occur once, but it's so fast you could actually run it every frame. Here’s the full pipeline: - Mesh-to-splat conversion pass - Depth pre-pass - Sort - Gaussian pre-pass (projective computations, etc.) - GBuffer rendering pass - Shadow pass - Deferred lighting pass This is the same classic deferred pipeline, but with Gaussians instead of triangles. Real-time relighting of converted assets? Totally possible.

✨ From Mesh to 3D Gaussians - Shaded in Deferred Style ✨ By Stefano Scolari (with link to repo in the comment) When converting a mesh to 3DGS model, the Gaussians are embedded with PBR properties and normal info during conversion. This prepares them for shading, similar to traditional geometry. In this scenario, Stefano used a deferred approach for shading to demonstrate how seamlessly it translates from triangle-based pipelines. 🔥 Fun fact: The mesh-to-splat conversion only needs to occur once, but it's so fast you could actually run it every frame. Here’s the full pipeline: - Mesh-to-splat conversion pass - Depth pre-pass - Sort - Gaussian pre-pass (projective computations, etc.) - GBuffer rendering pass - Shadow pass - Deferred lighting pass This is the same classic deferred pipeline, but with Gaussians instead of triangles. Real-time relighting of converted assets? Totally possible.

MrNeRF

19,768 views • 1 year ago

This is the Canadian model and it has been consistently cranking out a solid snow for Kentucky by Sunday morning. *IF* this ends up close to the final outcome it will be one of the greatest wins versus other models. This needs to be monitored!!!

This is the Canadian model and it has been consistently cranking out a solid snow for Kentucky by Sunday morning. IF this ends up close to the final outcome it will be one of the greatest wins versus other models. This needs to be monitored!!!

Jude Redfield

117,611 views • 1 year ago

seeing people say AI UGC is worse than human UGC let’s do the real comparison Human UGC: slow, inconsistent, expensive = shipping, 7–14 day waits, bad audio, missed scripts, revisions, delays AI UGC: instant, unlimited, zero headache = 50+ variations in minutes, perfect lighting, endless hooks, consistent delivery scaling ads isn’t about cost per video it’s about testing speed AI finds winners faster every time Comment “UGC” and I’ll show you how to do it. (must be following)

seeing people say AI UGC is worse than human UGC let’s do the real comparison Human UGC: slow, inconsistent, expensive = shipping, 7–14 day waits, bad audio, missed scripts, revisions, delays AI UGC: instant, unlimited, zero headache = 50+ variations in minutes, perfect lighting, endless hooks, consistent delivery scaling ads isn’t about cost per video it’s about testing speed AI finds winners faster every time Comment “UGC” and I’ll show you how to do it. (must be following)

Cas.Fyn

15,064 views • 7 months ago

That single white streak is an interstellar rock moving faster than anything humanity has ever launched. It didn’t strike Saturn itself—it slammed straight into the rings. Watch closely over the next eight seconds. The gap opening before your eyes will never close. It’s permanent. Saturn’s rings are disappearing in real time, and there’s nothing anyone can do to stop it.

That single white streak is an interstellar rock moving faster than anything humanity has ever launched. It didn’t strike Saturn itself—it slammed straight into the rings. Watch closely over the next eight seconds. The gap opening before your eyes will never close. It’s permanent. Saturn’s rings are disappearing in real time, and there’s nothing anyone can do to stop it.

Physics & Astronomy Zone

24,664 views • 2 months ago

Who's most likely to catch a spider and release it? L: *holds up oscar* he grew up with them! O: I mean, neither of us! I've never seen lando have to react to a spider...if it's the same as fish it would be funny! IM CRYING 😭😭😭

Who's most likely to catch a spider and release it? L: holds up oscar he grew up with them! O: I mean, neither of us! I've never seen lando have to react to a spider...if it's the same as fish it would be funny! IM CRYING 😭😭😭

a⁸¹ 🪼 | OSCAR PIASTRI 7X RACE WINNER

91,930 views • 1 year ago

NVIDIA just made AI detect objects 10x faster by deleting one step. It's called LocateAnything, and it removes the biggest bottleneck no one else was fixing in vision-language models. Normally a model builds each bounding box one coordinate token at a time. 100 objects means thousands of tokens before an answer. NVIDIA scrapped that: their Parallel Box Decoding predicts the whole box in a single forward pass, as one atomic unit. → 12.7 boxes/sec on one H100 → 10x faster than Qwen3-VL → +3.8% F1 on LVIS, accuracy up, not down → 3B params, runs on one consumer GPU Treating the box as one unit keeps its coordinates tied together, which is why accuracy climbed instead of falling. One model handles detection, GUI grounding, OCR, and document understanding, ready for computer-use agents, robotics, and document pipelines. 100% open source, weights, code, demo, and paper all live.

NVIDIA just made AI detect objects 10x faster by deleting one step. It's called LocateAnything, and it removes the biggest bottleneck no one else was fixing in vision-language models. Normally a model builds each bounding box one coordinate token at a time. 100 objects means thousands of tokens before an answer. NVIDIA scrapped that: their Parallel Box Decoding predicts the whole box in a single forward pass, as one atomic unit. → 12.7 boxes/sec on one H100 → 10x faster than Qwen3-VL → +3.8% F1 on LVIS, accuracy up, not down → 3B params, runs on one consumer GPU Treating the box as one unit keeps its coordinates tied together, which is why accuracy climbed instead of falling. One model handles detection, GUI grounding, OCR, and document understanding, ready for computer-use agents, robotics, and document pipelines. 100% open source, weights, code, demo, and paper all live.

Alvaro Cintas

201,044 views • 29 days ago

Something I just have to point out during my editing today Its pretty amazing how you can train a single magnetic mask. One per eye. And then in the streamed SBS A/V output: it shows that accuracy in real time It actually is easy to be green!

Something I just have to point out during my editing today Its pretty amazing how you can train a single magnetic mask. One per eye. And then in the streamed SBS A/V output: it shows that accuracy in real time It actually is easy to be green!

Brad Lynch

14,705 views • 5 months ago

I'm so happy that PhoneBuff mentioned this extremely important fact about battery tests And it's not just Apples internal testing. It's 9/10 battery tests on YouTube. They never touch the screen once which DOES in fact has a huge impact on battery life Every single time you touch your phones screen, it has to ramp up the processor to not just calculate the touch input, it hast to calculate predictably since you may not just be touching, but swiping. And it should continue scrolling once you lift your finger because that makes it feel more natural. And this needs power. Drains power. So a more important touchscreen and software can make a huge difference in battery life. And most battery tests just ignore that entirely. That's why some tests show vastly different results

I'm so happy that PhoneBuff mentioned this extremely important fact about battery tests And it's not just Apples internal testing. It's 9/10 battery tests on YouTube. They never touch the screen once which DOES in fact has a huge impact on battery life Every single time you touch your phones screen, it has to ramp up the processor to not just calculate the touch input, it hast to calculate predictably since you may not just be touching, but swiping. And it should continue scrolling once you lift your finger because that makes it feel more natural. And this needs power. Drains power. So a more important touchscreen and software can make a huge difference in battery life. And most battery tests just ignore that entirely. That's why some tests show vastly different results

Tim 🏳️‍🌈

71,908 views • 9 months ago

I just ran Gemma 4 31B on @CerebrasSystems at 1,800+ tokens/sec and it's multimodal. For context: that's 35x faster than a typical GPU endpoint, and the first token (reasoning included) lands in 1.5 seconds. This isn't a benchmark slide, I recorded the inference live. Prompt I used: "Create a simulation of an iPhone. Include at least one working dummy note taking app, a functional notification pulldown, high quality graphics, single HTML file, any libs via CDN." - Generation time: 3 seconds. - Notes app worked. - Notification panel worked. - Rendered first try. This is what wafer-scale inference unlocks, not just "faster," but a different category of product. When generation is this fast, you stop waiting and start iterating in real time. Why this matters: Gemma 4 31B is Google DeepMind's flagship open weight model, Apache 2.0 licensed, dense (not MoE), and built for efficiency over raw parameter count. It scores close to Claude Haiku 4.5 on the Artificial Analysis Intelligence Index (30 vs 29) but runs ~18x faster on Cerebras. It's also the first multimodal model on Cerebras's platform, meaning you can now feed it screenshots, documents, charts, and UI states at wafer scale speed. # Applications I'm most excited about: - Screenshot → Insight: Drop in a dashboard or document screenshot, get structured findings back instantly. no waiting, no batching. - Live UI generation: Full interactive interfaces (like my iPhone sim) generated and rendered in under 2 seconds. - Screenshot -> Patch: Feed it a broken UI + console error, get a minimal code fix and verification steps back. - Computer use & agentic loops: See -> reason -> act - verify, fast enough to keep a human in the loop instead of waiting on the model. - Long context summarization: Full research reports condensed into decision ready summaries you can read and requery in one sitting. The bigger unlock isn't the speed number itself, it's that agentic and multimodal loops (see -> reason -> output -> tool call -> verify -> retry) finally run in real time instead of feeling sluggish. As Logan Kilpatrick (Logan Kilpatrick) put it: "If every model was doing 2,000 tokens per second, you wouldn't build the same product and just have it be faster, you'd build different products." Gemma 4 31B is live now on Cerebras Inference Cloud in public preview. If you're building multimodal, agentic, or real time apps, this is worth testing today. What would you build with such insane inference throughput?

I just ran Gemma 4 31B on @CerebrasSystems at 1,800+ tokens/sec and it's multimodal. For context: that's 35x faster than a typical GPU endpoint, and the first token (reasoning included) lands in 1.5 seconds. This isn't a benchmark slide, I recorded the inference live. Prompt I used: "Create a simulation of an iPhone. Include at least one working dummy note taking app, a functional notification pulldown, high quality graphics, single HTML file, any libs via CDN." - Generation time: 3 seconds. - Notes app worked. - Notification panel worked. - Rendered first try. This is what wafer-scale inference unlocks, not just "faster," but a different category of product. When generation is this fast, you stop waiting and start iterating in real time. Why this matters: Gemma 4 31B is Google DeepMind's flagship open weight model, Apache 2.0 licensed, dense (not MoE), and built for efficiency over raw parameter count. It scores close to Claude Haiku 4.5 on the Artificial Analysis Intelligence Index (30 vs 29) but runs ~18x faster on Cerebras. It's also the first multimodal model on Cerebras's platform, meaning you can now feed it screenshots, documents, charts, and UI states at wafer scale speed. # Applications I'm most excited about: - Screenshot → Insight: Drop in a dashboard or document screenshot, get structured findings back instantly. no waiting, no batching. - Live UI generation: Full interactive interfaces (like my iPhone sim) generated and rendered in under 2 seconds. - Screenshot -> Patch: Feed it a broken UI + console error, get a minimal code fix and verification steps back. - Computer use & agentic loops: See -> reason -> act - verify, fast enough to keep a human in the loop instead of waiting on the model. - Long context summarization: Full research reports condensed into decision ready summaries you can read and requery in one sitting. The bigger unlock isn't the speed number itself, it's that agentic and multimodal loops (see -> reason -> output -> tool call -> verify -> retry) finally run in real time instead of feeling sluggish. As Logan Kilpatrick (Logan Kilpatrick) put it: "If every model was doing 2,000 tokens per second, you wouldn't build the same product and just have it be faster, you'd build different products." Gemma 4 31B is live now on Cerebras Inference Cloud in public preview. If you're building multimodal, agentic, or real time apps, this is worth testing today. What would you build with such insane inference throughput?

Alok

12,962 views • 26 days ago

Day 12/90 of Inference Engineering What is chunked prefill within vLLM? In continuation of yesterday's post on the high level architecture of vLLM, I want to dive deeper into vLLM core engine starting with the mechanics of chunked prefill. In this post, I will closely follow the original blog on the anatomy of vLLM. To start, let's define chunked prefill. It's a runtime inference optimization technique that splits a long input request so that it doesn’t monopolize the whole GPU. Keep in mind this is all within the context of vLLM. And since vLLM is an inference engine that's meant to serve a model to multiple concurrent users, having a GPU that’s fully monopolized on a single user's request means other users' requests would be in queue waiting to be processed. It isn’t too good to have the whole GPU occupied on a single request when the GPU is meant to be shared! So the key idea behind chunked prefill is to break the long request into smaller chunks, so that each chunk along with other users' requests gets processed and written into the KV cache together. Suppose we split up the long request into chunks and each chunk has 8 tokens. Now each memory block can hold 4 tokens. Therefore, 8 tokens can fit into 2 blocks of memory. After the first forward pass, 2 blocks are occupied, and after the second forward pass, 4 blocks of memory are occupied and so forth. Each forward pass handles a small chunk of the long request so that there's room in the same pass to keep serving other users' requests. Here's a small animation that I made today to fully visualize the idea behind chunked prefill when learning this topic~

Day 12/90 of Inference Engineering What is chunked prefill within vLLM? In continuation of yesterday's post on the high level architecture of vLLM, I want to dive deeper into vLLM core engine starting with the mechanics of chunked prefill. In this post, I will closely follow the original blog on the anatomy of vLLM. To start, let's define chunked prefill. It's a runtime inference optimization technique that splits a long input request so that it doesn’t monopolize the whole GPU. Keep in mind this is all within the context of vLLM. And since vLLM is an inference engine that's meant to serve a model to multiple concurrent users, having a GPU that’s fully monopolized on a single user's request means other users' requests would be in queue waiting to be processed. It isn’t too good to have the whole GPU occupied on a single request when the GPU is meant to be shared! So the key idea behind chunked prefill is to break the long request into smaller chunks, so that each chunk along with other users' requests gets processed and written into the KV cache together. Suppose we split up the long request into chunks and each chunk has 8 tokens. Now each memory block can hold 4 tokens. Therefore, 8 tokens can fit into 2 blocks of memory. After the first forward pass, 2 blocks are occupied, and after the second forward pass, 4 blocks of memory are occupied and so forth. Each forward pass handles a small chunk of the long request so that there's room in the same pass to keep serving other users' requests. Here's a small animation that I made today to fully visualize the idea behind chunked prefill when learning this topic~

max fu

28,891 views • 8 days ago