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。

today we are teaching a machine to exploit object detection models using swarm drones + reinforcement learning after only 5,000 steps on my Mac Air M3, the policy brought down detection from 90% to 32%, allowing 20+ drones to fly undetected

chester

15,736 subscribers

180,750 次观看 • 1 年前 •via X (Twitter)

健康养生科学技术体育

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

we are introducing a state-of-the-art real-time object detection model, RF-DETR RF-DETR outperforms all existing object detection models on real world datasets and is the first real-time model to achieve 60+ Average Precision on COCO talked w/ NVIDIA about it at GTC:

we are introducing a state-of-the-art real-time object detection model, RF-DETR RF-DETR outperforms all existing object detection models on real world datasets and is the first real-time model to achieve 60+ Average Precision on COCO talked w/ NVIDIA about it at GTC:

Joseph Nelson

18,231 次观看 • 1 年前

i gave a voice to my jarvis system everything running in real-time using open source models - roboflow RF-DETR for object detection - gemma 4 for scene summarization - kokoro text-to-speech - live in the browser using transformers js prompt: "you are a dystopian science fiction narrator in a cyberpunk world. describe the scene in one sentence"

i gave a voice to my jarvis system everything running in real-time using open source models - roboflow RF-DETR for object detection - gemma 4 for scene summarization - kokoro text-to-speech - live in the browser using transformers js prompt: "you are a dystopian science fiction narrator in a cyberpunk world. describe the scene in one sentence"

AA

28,753 次观看 • 5 个月前

Even though The Bullfrog, Allen Control Systems's AI-powered machine gun, is designed to shoot down Group 1 and 2 drones, which are by far the most difficult to destroy, it can still damage Group 3 Iranian Shahed drones. "We are currently focused on being the best at shooting down Group 1 and Group 2 drones, which are absolutely the hardest drones to engage... That said, we can shoot down Group 1 drones that are very far away." - CEO Michael Wior "So if I can hit a 7-inch target on a Group 1 drone from very far away, I can hit very vulnerable parts of a [Group 3] Shahed at the same distance." "Lasers are pretty good at shooting down Group 3 drones, but you have to keep your laser on target for 5 to 15 seconds in order to damage the airframe." "So while we're not the ideal solution for engaging those larger targets, when they get closer and you need that last line of defense, we are absolutely a great solution for taking those out of the air."

Even though The Bullfrog, Allen Control Systems's AI-powered machine gun, is designed to shoot down Group 1 and 2 drones, which are by far the most difficult to destroy, it can still damage Group 3 Iranian Shahed drones. "We are currently focused on being the best at shooting down Group 1 and Group 2 drones, which are absolutely the hardest drones to engage... That said, we can shoot down Group 1 drones that are very far away." - CEO Michael Wior "So if I can hit a 7-inch target on a Group 1 drone from very far away, I can hit very vulnerable parts of a [Group 3] Shahed at the same distance." "Lasers are pretty good at shooting down Group 3 drones, but you have to keep your laser on target for 5 to 15 seconds in order to damage the airframe." "So while we're not the ideal solution for engaging those larger targets, when they get closer and you need that last line of defense, we are absolutely a great solution for taking those out of the air."

TBPN

15,033 次观看 • 2 个月前

The network for machine intelligence Two years ago, we laid out our vision for a machine learning compute protocol. One that connects every device in the world into an open network for machine intelligence, with no gatekeepers or artificial boundaries. This week, we’ll be sharing some of our early progress, beginning with RL Swarm, a peer-to-peer system for collaborative reinforcement learning over the internet. Next month, we’ll open our Testnet, allowing anyone to contribute to the frontier of open machine intelligence. Introducing RL Swarm RL Swarm is a fully open source system for collaborative reinforcement learning over the internet. It is a live demo of our research findings, which show that models training with RL learn faster when they train as a collective swarm than they do on their own. Join our swarm now to see this in practice. You can participate with consumer hardware at home or a powerful GPU in the cloud. You can follow along with the swarm’s progress by following the links below.

The network for machine intelligence Two years ago, we laid out our vision for a machine learning compute protocol. One that connects every device in the world into an open network for machine intelligence, with no gatekeepers or artificial boundaries. This week, we’ll be sharing some of our early progress, beginning with RL Swarm, a peer-to-peer system for collaborative reinforcement learning over the internet. Next month, we’ll open our Testnet, allowing anyone to contribute to the frontier of open machine intelligence. Introducing RL Swarm RL Swarm is a fully open source system for collaborative reinforcement learning over the internet. It is a live demo of our research findings, which show that models training with RL learn faster when they train as a collective swarm than they do on their own. Join our swarm now to see this in practice. You can participate with consumer hardware at home or a powerful GPU in the cloud. You can follow along with the swarm’s progress by following the links below.

gensyn

228,915 次观看 • 1 年前

#WATCH | Ambala, Haryana | Major Varun Jeet Singh says, "We have developed this system and brought it from Assam... There is a detection system that identifies threats, allowing us to neutralise them... The jammer first tries to jam the frequencies of the enemy's drone. If missed, there is a hard kill option that fires shots on the drones... If it still misses, a rocket will go up by 100-150 meters and finish any possible drone..."

#WATCH | Ambala, Haryana | Major Varun Jeet Singh says, "We have developed this system and brought it from Assam... There is a detection system that identifies threats, allowing us to neutralise them... The jammer first tries to jam the frequencies of the enemy's drone. If missed, there is a hard kill option that fires shots on the drones... If it still misses, a rocket will go up by 100-150 meters and finish any possible drone..."

ANI

144,586 次观看 • 11 个月前

Zelensky: It is important that airlines using Russian airspace continue to pay attention to the real situation and the real danger. Today, there are new signals from airlines on this, and we confirm that because of Russia’s war, Russian skies have become a space for drones, not civilian aircraft. Drones are a concrete reality, unlike the words of Russian officials who are trying to reassure airlines.

Zelensky: It is important that airlines using Russian airspace continue to pay attention to the real situation and the real danger. Today, there are new signals from airlines on this, and we confirm that because of Russia’s war, Russian skies have become a space for drones, not civilian aircraft. Drones are a concrete reality, unlike the words of Russian officials who are trying to reassure airlines.

Clash Report

14,571 次观看 • 3 天前

More strikes by Ukrainian kamikaze drones on the Russian Rakushechnoye and Yuri Korchagin drilling platforms in the Caspian Sea, more than 1,000 km from the line of contact. The enemy drones were presumably controlled via Starlink. It can be seen that the platforms were equipped with light air defense systems or man-portable air defense systems to protect them, which they used to attempt to shoot down the drones, but failed to repel the attack. Military Informant Ukrainian drones have again attacked targets in Russia at a depth of 1,000 km, using Starlink. Elon Musk authorized these attacks. When will there be a response? We need to shoot down all his fucking satellites.

More strikes by Ukrainian kamikaze drones on the Russian Rakushechnoye and Yuri Korchagin drilling platforms in the Caspian Sea, more than 1,000 km from the line of contact. The enemy drones were presumably controlled via Starlink. It can be seen that the platforms were equipped with light air defense systems or man-portable air defense systems to protect them, which they used to attempt to shoot down the drones, but failed to repel the attack. Military Informant Ukrainian drones have again attacked targets in Russia at a depth of 1,000 km, using Starlink. Elon Musk authorized these attacks. When will there be a response? We need to shoot down all his fucking satellites.

Victor vicktop55 commentary

12,506 次观看 • 4 个月前

🔥 Ukraine shoots down Russian “Geran-3” drones with a new interceptor — first footage The debut of the Ukrainian-made Sting drone proved to be highly successful. On November 30, the Stings destroyed several Russian jet-powered “Geran-3” drones. These drones are faster than the regular “Geran” models and more maneuverable, designed to break through air defenses. Russia has been using “Geran-3” in strike packages since the summer of 2025. Now Ukraine has a tool capable of disrupting this tactic.

🔥 Ukraine shoots down Russian “Geran-3” drones with a new interceptor — first footage The debut of the Ukrainian-made Sting drone proved to be highly successful. On November 30, the Stings destroyed several Russian jet-powered “Geran-3” drones. These drones are faster than the regular “Geran” models and more maneuverable, designed to break through air defenses. Russia has been using “Geran-3” in strike packages since the summer of 2025. Now Ukraine has a tool capable of disrupting this tactic.

NEXTA

63,378 次观看 • 9 个月前

basically, a huge % of the views you’re paying for are FAKE this indian botting group is literally teaching people how to professionally bot “content rewards” clipping platform cpm-based clipping without REAL bot detection is a SCAM period. not having bot detection creates a massive opportunity for blackhats to build entire business models around exploiting these platforms and they’re not doing it sloppily anymore they bot professionally fake engagement today is sophisticated enough that the human eye will never reliably detect it that’s exactly why we spent years and millions of dollars building bot detection systems for this space and honestly, we’re the only platform that actually solved it

basically, a huge % of the views you’re paying for are FAKE this indian botting group is literally teaching people how to professionally bot “content rewards” clipping platform cpm-based clipping without REAL bot detection is a SCAM period. not having bot detection creates a massive opportunity for blackhats to build entire business models around exploiting these platforms and they’re not doing it sloppily anymore they bot professionally fake engagement today is sophisticated enough that the human eye will never reliably detect it that’s exactly why we spent years and millions of dollars building bot detection systems for this space and honestly, we’re the only platform that actually solved it

f*ckgrowth

11,225 次观看 • 3 个月前

Turkiyé and China are the only true friends Pakistan has left. It is their drones that Pakistan is attempting to fly in swarms across the border, I think, to probe for any gaps in our Air Defence grid, which had proved to be very robust. Hear my take on WION

Turkiyé and China are the only true friends Pakistan has left. It is their drones that Pakistan is attempting to fly in swarms across the border, I think, to probe for any gaps in our Air Defence grid, which had proved to be very robust. Hear my take on WION

Lt Gen Satish Dua 🇮🇳

248,640 次观看 • 1 年前

I am still very surprised there has not been an effort to search the desert areas for #NancyGuthrie. I previously showed an example of using drone still images to map entire areas and then applying specialized detection software. This is another example, using video captured from a drone and detection software programmed to identify humans and human remains. In this case, from a few months ago, I located this individual using that method. The deceased person was wearing clothing that blended into the environment and was largely concealed under brush, yet we were still able to detect the person while covering large areas of land from high altitudes with drones and specialized software. I feel the Sheriff’s Department and/or the family should consider this type of strategy in the areas with the highest probability of detection. #SAR

I am still very surprised there has not been an effort to search the desert areas for #NancyGuthrie. I previously showed an example of using drone still images to map entire areas and then applying specialized detection software. This is another example, using video captured from a drone and detection software programmed to identify humans and human remains. In this case, from a few months ago, I located this individual using that method. The deceased person was wearing clothing that blended into the environment and was largely concealed under brush, yet we were still able to detect the person while covering large areas of land from high altitudes with drones and specialized software. I feel the Sheriff’s Department and/or the family should consider this type of strategy in the areas with the highest probability of detection. #SAR

SF INVESTIGATES • STEVE FISCHER

56,610 次观看 • 5 个月前

We started with simple kamikaze sea drones. Then we built drones with turrets that can shoot down helicopters. Now we have drones that can shoot down Russian fighter jets from the sea. We have developed boats that carry other drones. We also have boats that strike targets on land from the sea. And we are developing more stable drones that can operate longer and more effectively at sea. Soon – and not in the distant future – we will have systems that can operate even in ocean conditions. We are also actively working on underwater systems. So, facing threats in the Black Sea, we are finding the right security solutions. These solutions can be used in your matters as well – including in complex situations like the Strait of Hormuz today. Drones can solve problems that even a fleet sometimes cannot. From my address to the Parliament of the United Kingdom (6/7)

We started with simple kamikaze sea drones. Then we built drones with turrets that can shoot down helicopters. Now we have drones that can shoot down Russian fighter jets from the sea. We have developed boats that carry other drones. We also have boats that strike targets on land from the sea. And we are developing more stable drones that can operate longer and more effectively at sea. Soon – and not in the distant future – we will have systems that can operate even in ocean conditions. We are also actively working on underwater systems. So, facing threats in the Black Sea, we are finding the right security solutions. These solutions can be used in your matters as well – including in complex situations like the Strait of Hormuz today. Drones can solve problems that even a fleet sometimes cannot. From my address to the Parliament of the United Kingdom (6/7)

Volodymyr Zelenskyy / Володимир Зеленський

343,317 次观看 • 5 个月前

Today we introduce iBuild, the revolutionary platform allowing anyone to build a dApp, from start to finish using only natural language prompts. dApp development will never be the same. iBuild - only on Injective 🥷. The future starts here.

Today we introduce iBuild, the revolutionary platform allowing anyone to build a dApp, from start to finish using only natural language prompts. dApp development will never be the same. iBuild - only on Injective 🥷. The future starts here.

Injective 🥷

58,206 次观看 • 1 年前

Militants in Syria have wrapped themselves in blue and yellow tape, either after seeing it from the Ukrainian Armed Forces or on someone's advice. Now the militants are trying to break through to Aleppo using drones and armored vehicles. The Syrian Air Force and Russian Aerospace Forces are working on them. - AN

Militants in Syria have wrapped themselves in blue and yellow tape, either after seeing it from the Ukrainian Armed Forces or on someone's advice. Now the militants are trying to break through to Aleppo using drones and armored vehicles. The Syrian Air Force and Russian Aerospace Forces are working on them. - AN

Zlatti71

300,455 次观看 • 1 年前

🇪🇺🦅 Europe is arming itself against Russian drones, including FPVs that attack or monitor infrastructure or equipment. A Swedish firm has created a cheap mini-missile to destroy UAVs, – The Telegraph – Hunts drones using infrared sensors. – Destroys UAVs with a blow, not an explosive — so there are no shrapnel and the risk of fire. – Finally, Europe will not have to lift fighter jets into the air — these mini-missiles can be used, because they are 16,000 times cheaper to operate. – Other advantages: a range of more than 3 km, autonomous detection and tracking using built-in AI. – The missile can be launched from a hand cannon, a rifle-type device or a vehicle.

🇪🇺🦅 Europe is arming itself against Russian drones, including FPVs that attack or monitor infrastructure or equipment. A Swedish firm has created a cheap mini-missile to destroy UAVs, – The Telegraph – Hunts drones using infrared sensors. – Destroys UAVs with a blow, not an explosive — so there are no shrapnel and the risk of fire. – Finally, Europe will not have to lift fighter jets into the air — these mini-missiles can be used, because they are 16,000 times cheaper to operate. – Other advantages: a range of more than 3 km, autonomous detection and tracking using built-in AI. – The missile can be launched from a hand cannon, a rifle-type device or a vehicle.

MAKS 25 🇺🇦👀

45,874 次观看 • 11 个月前

Russian drones consist of 90% Chinese components. Hackers gained access to a closed-door meeting of the Russian Ministry of Industry and Trade, where the issue of dependency on China was discussed. It is alleged that over 90% of the key components for Russian drones are supplied from the North Korea. Russian companies do not produce their own engines, electronics, plastic, or batteries, they merely assemble drones using imported components.

Russian drones consist of 90% Chinese components. Hackers gained access to a closed-door meeting of the Russian Ministry of Industry and Trade, where the issue of dependency on China was discussed. It is alleged that over 90% of the key components for Russian drones are supplied from the North Korea. Russian companies do not produce their own engines, electronics, plastic, or batteries, they merely assemble drones using imported components.

Anton Gerashchenko

35,226 次观看 • 4 个月前

i made a computer vision system to monitor the situation - live webcam feed and roboflow RF-DETR for object detection - inputs are fed into SmolVLM which describes what it sees in one sentence - dataviz with vanilla js all running locally in real-time on my macbook air m2

i made a computer vision system to monitor the situation - live webcam feed and roboflow RF-DETR for object detection - inputs are fed into SmolVLM which describes what it sees in one sentence - dataviz with vanilla js all running locally in real-time on my macbook air m2

AA

72,555 次观看 • 6 个月前

.Trenton Bricken explains how we know LLMs are actually generalizing - aka they're not just stochastic parrots: - Training models on code makes them better at reasoning in language. - Models fine tuned on math problems become better at entity detection. - We can just straightforwardly read the world-models developed by smaller NNs which are easier to interpret (Othello). Transfer learning shows models are developing a deeper understanding of their data. Full episode out Thursday.

.Trenton Bricken explains how we know LLMs are actually generalizing - aka they're not just stochastic parrots: - Training models on code makes them better at reasoning in language. - Models fine tuned on math problems become better at entity detection. - We can just straightforwardly read the world-models developed by smaller NNs which are easier to interpret (Othello). Transfer learning shows models are developing a deeper understanding of their data. Full episode out Thursday.

Dwarkesh Patel

237,812 次观看 • 2 年前

Check out our #ECCV2026 paper "Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention", where we make linear attention sparse in space, recurrent in time, and parallel in training, enabling the first purely-linear-attention-based neural network for asynchronous object detection with #EventCameras, outperforming the previous best asynchronous method with 20x less computation with truly event-by-event inference on CPU! Code released! Paper: Code: Video: Event cameras promise extremely low-latency vision, but to fully exploit them, the neural network must be low-latency too. We introduce #SpatiallySparseLinearAttention (#SSLA) for asynchronous object detection directly from raw events. Linear attention is particularly appealing for event cameras: it can be trained efficiently in parallel on long event sequences, while at inference it operates recurrently, updating its prediction every time a new event arrives. The problem is that conventional linear attention updates its entire state for every event. For object detection, where fine spatial resolution matters, this quickly becomes expensive. Our key idea is simple: an event only carries information about a small spatial region, so why update the entire spatial state? SSLA updates only the relevant parts of the state, enabling fine-grained spatial representations while keeping per-event computation low. We achieve: - >20× lower per-event computation than the strongest prior asynchronous baseline - State-of-the-art accuracy among asynchronous object detection methods - Truly event-by-event inference on CPU, designed to preserve the latency advantage of event cameras Come to our poster on Friday September 11, 2026 from 4-6pm at ExHall #389 Reference: Haiqing Hao, Zhipeng Sui, Rong Zou, Zijia Dai, Nikola Zubić, Davide Scaramuzza, Wenhui Wang Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention ECCV, 2026 Prophesee SynSense University of Zurich UZH Science European Research Council (ERC) UZHai UZH IfI Tesla BYD #EventCameras #ComputerVision #Robotics #DeepLearning #NeuromorphicVision #AI

Check out our #ECCV2026 paper "Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention", where we make linear attention sparse in space, recurrent in time, and parallel in training, enabling the first purely-linear-attention-based neural network for asynchronous object detection with #EventCameras, outperforming the previous best asynchronous method with 20x less computation with truly event-by-event inference on CPU! Code released! Paper: Code: Video: Event cameras promise extremely low-latency vision, but to fully exploit them, the neural network must be low-latency too. We introduce #SpatiallySparseLinearAttention (#SSLA) for asynchronous object detection directly from raw events. Linear attention is particularly appealing for event cameras: it can be trained efficiently in parallel on long event sequences, while at inference it operates recurrently, updating its prediction every time a new event arrives. The problem is that conventional linear attention updates its entire state for every event. For object detection, where fine spatial resolution matters, this quickly becomes expensive. Our key idea is simple: an event only carries information about a small spatial region, so why update the entire spatial state? SSLA updates only the relevant parts of the state, enabling fine-grained spatial representations while keeping per-event computation low. We achieve: - >20× lower per-event computation than the strongest prior asynchronous baseline - State-of-the-art accuracy among asynchronous object detection methods - Truly event-by-event inference on CPU, designed to preserve the latency advantage of event cameras Come to our poster on Friday September 11, 2026 from 4-6pm at ExHall #389 Reference: Haiqing Hao, Zhipeng Sui, Rong Zou, Zijia Dai, Nikola Zubić, Davide Scaramuzza, Wenhui Wang Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention ECCV, 2026 Prophesee SynSense University of Zurich UZH Science European Research Council (ERC) UZHai UZH IfI Tesla BYD #EventCameras #ComputerVision #Robotics #DeepLearning #NeuromorphicVision #AI

Davide Scaramuzza

51,571 次观看 • 11 天前