正在加载视频...

视频加载失败

返回首页

Miniscope Zero, the first fully wireless Miniscope, is on bioRxiv! Wireless power + data enable neural recordings in enclosed mazes, 3D environments & from two animals at once! Dr. Marcel Brosch Miniscope Team Paper: Wiki:

takuyasasatani's profile picture

Takuya SASATANI

1,048 subscribers

23,293 次观看 • 14 天前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

6 条评论

Takuya SASATANI 的头像
Takuya SASATANI14 天前

Team: T. Sasatani & M. Brosch @BroschMarcel (co-first) Z. Dong, J. Saunders @json_dirs, P. Zhao @PingpingION, M. Sehgal @MeghaSehgal_, L. Oesch, F.S. Jimka, H. Semwal, A. Guha, H. Chorsi, K. Keus @karina_keus, B. Madruga @BlakeMadruga, A. Silva, A.K. Churchland @anne_churchland, P. Golshani @pgolshani & D. Aharoni. @MiniscopeTeam

Isamu Amir 的头像
Isamu Amir14 天前

@biorxivpreprint @BroschMarcel @MiniscopeTeam おめでとうございます！素晴らしい！

Takuya SASATANI 的头像
Takuya SASATANI14 天前

@biorxivpreprint @BroschMarcel @MiniscopeTeam ありがとうございます！

Burak Gençay 的头像
Burak Gençay14 天前

@biorxivpreprint @BroschMarcel @MiniscopeTeam Dear Takuya, I would love to work with you and your team.

eat organic, don't oil panic 的头像
eat organic, don't oil panic14 天前

@biorxivpreprint @BroschMarcel @MiniscopeTeam Poor things.

Heikki Kallio 的头像
Heikki Kallio14 天前

@biorxivpreprint @BroschMarcel @MiniscopeTeam @grok what kind of data could we gather from here that could be animated as easily understandable images, and/or writing?

相关视频

We're presenting Miniscope Zero, the first fully-wireless Miniscope, at #SfN25! Wireless power + data enable long-term neural recordings from multiple freely behaving animals. Miniscope Team Poster: Nov. 17, 1–5 pm, ZZ7, Video:
1:42

We're presenting Miniscope Zero, the first fully-wireless Miniscope, at #SfN25! Wireless power + data enable long-term neural recordings from multiple freely behaving animals. Miniscope Team Poster: Nov. 17, 1–5 pm, ZZ7, Video:

Takuya SASATANI

15,092 次观看 • 10 个月前

We're presenting the first fully-wireless Miniscope at #SfN2023! Wireless power/data enables long-term neural recording! Miniscope Team Stop by our poster on Nov. 13, 8-12am, WW66. Untethering open-source Miniscopes with wireless power and data transfer
1:40

We're presenting the first fully-wireless Miniscope at #SfN2023! Wireless power/data enables long-term neural recording! Miniscope Team Stop by our poster on Nov. 13, 8-12am, WW66. Untethering open-source Miniscopes with wireless power and data transfer

Takuya SASATANI

61,976 次观看 • 2 年前

The SPRAWL zero demo is OUT NOW! We're thrilled you can finally get your hands on the demo on Steam now! To let you fully dive in, we’ve included the first two levels: The Graveyard and Buried. So much hard work from everyone here at MAETH, building the game of our dreams!
1:20

The SPRAWL zero demo is OUT NOW! We're thrilled you can finally get your hands on the demo on Steam now! To let you fully dive in, we’ve included the first two levels: The Graveyard and Buried. So much hard work from everyone here at MAETH, building the game of our dreams!

SPRAWL zero | OUT NOW!

271,184 次观看 • 3 个月前

Tulsa is up two scores at the half over Oklahoma State 😳 That's the Golden Hurricane's first double-digit halftime lead over a Power 4 team in the last 20 years 🤯 Watch the second half from Stillwater on ESPN and the ESPN App 👀
0:23

Tulsa is up two scores at the half over Oklahoma State 😳 That's the Golden Hurricane's first double-digit halftime lead over a Power 4 team in the last 20 years 🤯 Watch the second half from Stillwater on ESPN and the ESPN App 👀

ESPN Insights

380,751 次观看 • 1 年前

When we run out of good training data in reality, simulation is the next gold mine. Enters Infinigen: an open-source, procedurally generated, photorealistic dataset for 3D vision. The quality is stunning! No two worlds are the same. ▸ Every little detail is randomized and customizable, even the wrinkles on a flower petal! ▸ Diverse objects & scenes in the natural world: plants, animals, terrains; fire, cloud, rain, and snow. ▸ Groundtruth automatic annotations: optical flow, 3D scene flow, depth, surface normals, panoptic segmentation, occlusion boundaries. ▸ Written in Blender. Made by Jia Deng's lab at Princeton: ▸ Website: ▸ Code: ▸ Paper:
3:03

When we run out of good training data in reality, simulation is the next gold mine. Enters Infinigen: an open-source, procedurally generated, photorealistic dataset for 3D vision. The quality is stunning! No two worlds are the same. ▸ Every little detail is randomized and customizable, even the wrinkles on a flower petal! ▸ Diverse objects & scenes in the natural world: plants, animals, terrains; fire, cloud, rain, and snow. ▸ Groundtruth automatic annotations: optical flow, 3D scene flow, depth, surface normals, panoptic segmentation, occlusion boundaries. ▸ Written in Blender. Made by Jia Deng's lab at Princeton: ▸ Website: ▸ Code: ▸ Paper:

Jim Fan

363,656 次观看 • 3 年前

Open-ended coding training data may no longer be the bottleneck: AI can scale open-ended tasks—and even outperform human-expert curation. FrontierCS team is releasing FrontierSmith: a system for synthesizing open-ended coding problems at scale. Starting from closed-ended coding tasks, FrontierSmith mutates, filters, and builds runnable optimization environments for long-horizon coding agents. In our experiments, FrontierSmith data trains stronger models than human-curated open-ended data on FrontierCS and ALE-bench. Blog: Paper: Code: Model:
0:56

Open-ended coding training data may no longer be the bottleneck: AI can scale open-ended tasks—and even outperform human-expert curation. FrontierCS team is releasing FrontierSmith: a system for synthesizing open-ended coding problems at scale. Starting from closed-ended coding tasks, FrontierSmith mutates, filters, and builds runnable optimization environments for long-horizon coding agents. In our experiments, FrontierSmith data trains stronger models than human-curated open-ended data on FrontierCS and ALE-bench. Blog: Paper: Code: Model:

Qiuyang Mang

95,747 次观看 • 4 个月前

Evaluating Neural Networks at the Speed of Light (with Light!). See live optical inference in the video below. Excited to share recent academic work on optical neural networks as a collection of computing elements embedded in the camera lens! These elements perform computation optically even before an image is captured, using the photons in the scene instead of GPU computation after the capture. We were able to achieve ImageNet classification more than two orders of magnitude faster than conventional neural networks on today's GPUs at almost no power consumption! To do this, we developed an array of metalenses that perform this computation on light from the scene. Project: Paper: Amazing collaboration with Kaixuan Wei, Xiao Li, Johannes Froech, Praneeth Chakravarthula, James Whitehead, Ethan Tseng , Arka Majumdar .
0:45

Evaluating Neural Networks at the Speed of Light (with Light!). See live optical inference in the video below. Excited to share recent academic work on optical neural networks as a collection of computing elements embedded in the camera lens! These elements perform computation optically even before an image is captured, using the photons in the scene instead of GPU computation after the capture. We were able to achieve ImageNet classification more than two orders of magnitude faster than conventional neural networks on today's GPUs at almost no power consumption! To do this, we developed an array of metalenses that perform this computation on light from the scene. Project: Paper: Amazing collaboration with Kaixuan Wei, Xiao Li, Johannes Froech, Praneeth Chakravarthula, James Whitehead, Ethan Tseng , Arka Majumdar .

Felix Heide

29,904 次观看 • 1 年前

Dominion and ES&S have been found to have utilized wireless technology involved in the 2020 election, communicating with foreign IP addresses and equipment. During the forensics examination of an ICX machine in Michigan, there were two IP addresses discovered in unallocated space on the hard drive of the Linux operating system. The existence of these IPs in unallocated space implies the ICX had previous communication with one or both of the IPs. The first IP address was: 120.125.201.101. This IP address is registered to Ministry of Education Computer Center located in Taipei, Taiwan. The second IP address was: 62.146.7.95. This IP address is registered to EDV-BV GmbH QSC Subkunde located in Nurenberg, Germany. The printed summary tape from the ES&S machines also indicate that the submission of the vote totals occurred using the wireless 4G modem. The manufacturer of the wireless 4G card used in the ES&S DS200 is a company named Telit. Telit is an internet of things company that has recently taken major investment from a Chinese investment fund that has ties to the Chinese Communist Party. No other country's compromised equipment should EVER be allowed to process, send, or receive ANY data from the United States, especially our elections.
1:12

Dominion and ES&S have been found to have utilized wireless technology involved in the 2020 election, communicating with foreign IP addresses and equipment. During the forensics examination of an ICX machine in Michigan, there were two IP addresses discovered in unallocated space on the hard drive of the Linux operating system. The existence of these IPs in unallocated space implies the ICX had previous communication with one or both of the IPs. The first IP address was: 120.125.201.101. This IP address is registered to Ministry of Education Computer Center located in Taipei, Taiwan. The second IP address was: 62.146.7.95. This IP address is registered to EDV-BV GmbH QSC Subkunde located in Nurenberg, Germany. The printed summary tape from the ES&S machines also indicate that the submission of the vote totals occurred using the wireless 4G modem. The manufacturer of the wireless 4G card used in the ES&S DS200 is a company named Telit. Telit is an internet of things company that has recently taken major investment from a Chinese investment fund that has ties to the Chinese Communist Party. No other country's compromised equipment should EVER be allowed to process, send, or receive ANY data from the United States, especially our elections.

The SCIF

24,318 次观看 • 3 个月前

6G: FROM FASTER INTERNET TO UNDER-SKIN SURVEILLANCE? President Trump just bragged in front of some of the biggest names in tech about being a “leader on 5G” and now heading into 6G. Even joking about using it to get a “deeper view into somebody’s skin.” For years, anyone who asked whether advanced wireless technology could become something far beyond faster internet was written off as a conspiracy theorist. Now the president himself is publicly floating language that sounds a lot like peering inside the human body. 6G is essentially 5G on steroids, operating at terahertz frequencies. Lab studies have already linked these frequencies to DNA damage in neural stem cells, genomic instability, and cell death, and yet this technology is planned to be deployed virtually everywhere. So the question is simple: are we building these networks to help the public, or to enable real-time monitoring, surveillance, and tracking of entire populations? Every technological advancement, once sold as convenience, has pushed civil liberties further back. Now, with the rise of data centers, Flock cameras, WiFi sensing, brain implant chips, and the AI infrastructure now being built on top of it all, many are rightfully concerned. Theodora Scarato, Director of the Wireless & EMF Program at Environmental Health Sciences, joins me to break down what this means and where it’s headed.
15:35

6G: FROM FASTER INTERNET TO UNDER-SKIN SURVEILLANCE? President Trump just bragged in front of some of the biggest names in tech about being a “leader on 5G” and now heading into 6G. Even joking about using it to get a “deeper view into somebody’s skin.” For years, anyone who asked whether advanced wireless technology could become something far beyond faster internet was written off as a conspiracy theorist. Now the president himself is publicly floating language that sounds a lot like peering inside the human body. 6G is essentially 5G on steroids, operating at terahertz frequencies. Lab studies have already linked these frequencies to DNA damage in neural stem cells, genomic instability, and cell death, and yet this technology is planned to be deployed virtually everywhere. So the question is simple: are we building these networks to help the public, or to enable real-time monitoring, surveillance, and tracking of entire populations? Every technological advancement, once sold as convenience, has pushed civil liberties further back. Now, with the rise of data centers, Flock cameras, WiFi sensing, brain implant chips, and the AI infrastructure now being built on top of it all, many are rightfully concerned. Theodora Scarato, Director of the Wireless & EMF Program at Environmental Health Sciences, joins me to break down what this means and where it’s headed.

Ben Swann

10,396 次观看 • 1 个月前

PARTNERSHIP: Traditional mobile networks are struggling to adapt to the digital-first world of today. Prices keep rising, coverage gets worse, and customer data is sold to the highest bidder. Meet World Mobile, the AirBNB of telco, and the first web3 mobile network, fueled by $WMTX, which enables its users to reclaim power over their connectivity. World Mobile is a decentralized wireless network in which all network infrastructure is owned and operated by its community, powered by a sharing economy i.e. operators earn a share of the revenue. When disaster struck North Carolina, following the impact of Hurricane Helene, World Mobile deployed its decentralised wireless technology to restore critical connectivity, empowering communities when traditional networks failed. Watch the full story unfold in “Connecting the Disconnected” and witness how web3 innovation brings hope in times of crisis. Know more on and tune in to their live show at 4:30 pm EST on Mario Nawfal. Disclaimer: This is a paid partnership, DYOR as this is not financial advice.
1:11

PARTNERSHIP: Traditional mobile networks are struggling to adapt to the digital-first world of today. Prices keep rising, coverage gets worse, and customer data is sold to the highest bidder. Meet World Mobile, the AirBNB of telco, and the first web3 mobile network, fueled by $WMTX, which enables its users to reclaim power over their connectivity. World Mobile is a decentralized wireless network in which all network infrastructure is owned and operated by its community, powered by a sharing economy i.e. operators earn a share of the revenue. When disaster struck North Carolina, following the impact of Hurricane Helene, World Mobile deployed its decentralised wireless technology to restore critical connectivity, empowering communities when traditional networks failed. Watch the full story unfold in “Connecting the Disconnected” and witness how web3 innovation brings hope in times of crisis. Know more on and tune in to their live show at 4:30 pm EST on Mario Nawfal. Disclaimer: This is a paid partnership, DYOR as this is not financial advice.

Mario Nawfal

85,061 次观看 • 1 年前

Domain Randomization (DR) is a key component of the data augmentation pipeline at Axis Robotics. By applying DR, we are able to scale verified, high-quality human trajectories by 10x to 100x. During training, we systematically introduce variances in environmental parameters. This prevents the model from relying on spurious visual correlations. The objective is to ensure the policy learns rather than overfitting. To demonstrate the necessity and effectiveness of this approach, we evaluated both DR and No-DR models on Task 74 (pour_water_into_mug). The empirical results show a definitive impact on real-world deployment reliability: integrating DR into the pipeline increased the success rate from 0% to 90% (Fig. 1). This divergence stems from how the respective policies process visual observations (Fig. 2). The baseline (No DR) model overfits to the static visual background. It essentially memorizes the poses from the training dataset but fails to generalize when subjected to the inevitable variances of real-world deployment. Consequently, it cannot execute the correct manipulation on the target object. Conversely, the DR-trained model learns to extract essential geometric features and physical constraints, filtering out superficial visual noise. This leads to significantly higher robustness in dynamic environments. The structural difference in execution is clearly reflected in the end-effector trajectory data: These real-world deployment recordings further illustrate this difference (Videos 1 and 2). Scaling Physical AI requires turning raw trajectory data into robust policies, and a rigorously engineered DR infrastructure is an essential bridge to close the Sim2Real gap.
0:20

Domain Randomization (DR) is a key component of the data augmentation pipeline at Axis Robotics. By applying DR, we are able to scale verified, high-quality human trajectories by 10x to 100x. During training, we systematically introduce variances in environmental parameters. This prevents the model from relying on spurious visual correlations. The objective is to ensure the policy learns rather than overfitting. To demonstrate the necessity and effectiveness of this approach, we evaluated both DR and No-DR models on Task 74 (pour_water_into_mug). The empirical results show a definitive impact on real-world deployment reliability: integrating DR into the pipeline increased the success rate from 0% to 90% (Fig. 1). This divergence stems from how the respective policies process visual observations (Fig. 2). The baseline (No DR) model overfits to the static visual background. It essentially memorizes the poses from the training dataset but fails to generalize when subjected to the inevitable variances of real-world deployment. Consequently, it cannot execute the correct manipulation on the target object. Conversely, the DR-trained model learns to extract essential geometric features and physical constraints, filtering out superficial visual noise. This leads to significantly higher robustness in dynamic environments. The structural difference in execution is clearly reflected in the end-effector trajectory data: These real-world deployment recordings further illustrate this difference (Videos 1 and 2). Scaling Physical AI requires turning raw trajectory data into robust policies, and a rigorously engineered DR infrastructure is an essential bridge to close the Sim2Real gap.

Axis Robotics

27,125 次观看 • 5 个月前

Rylan Griffen says he feels like KU never fully recovered from the 2 OT loss to Houston at home back in January Added if the Houston game and blown lead at Baylor had gone differently, the season may have gone differently. But, he says things can't be blamed on just two games "This is the price you get when you get one of the lower seeds, middle-of-the-pack seeds .. you play against a another power five, high major team that was in a power conference. We did it to ourselves"
1:04

Rylan Griffen says he feels like KU never fully recovered from the 2 OT loss to Houston at home back in January Added if the Houston game and blown lead at Baylor had gone differently, the season may have gone differently. But, he says things can't be blamed on just two games "This is the price you get when you get one of the lower seeds, middle-of-the-pack seeds .. you play against a another power five, high major team that was in a power conference. We did it to ourselves"

Caroline Soro

405,412 次观看 • 1 年前

Starting the new year without human labeling 🎉!! Multimodal lidar-camera data is a gold mine of dense 3D geometry hiding in plain sight. For supervised pretraining and validation at scale at Torc-Robotics, we rely on fully automated pseudo-labeling pipelines. Exploiting geometric priors from temporally accumulated LiDAR maps and an iterative update rule enforces joint geometric–semantic consistency while detecting moving objects via inconsistencies. We achieve 3D semantic labels and 3D bounding boxes with human-like quality at 200m+ range required for highway driving. Paper: Exciting work with Torc-Robotics with Filippo Ghilotti, Samuel Brucker, Nahku Saidy, Matteo Matteucci, Mario Bijelic.
0:21

Starting the new year without human labeling 🎉!! Multimodal lidar-camera data is a gold mine of dense 3D geometry hiding in plain sight. For supervised pretraining and validation at scale at Torc-Robotics, we rely on fully automated pseudo-labeling pipelines. Exploiting geometric priors from temporally accumulated LiDAR maps and an iterative update rule enforces joint geometric–semantic consistency while detecting moving objects via inconsistencies. We achieve 3D semantic labels and 3D bounding boxes with human-like quality at 200m+ range required for highway driving. Paper: Exciting work with Torc-Robotics with Filippo Ghilotti, Samuel Brucker, Nahku Saidy, Matteo Matteucci, Mario Bijelic.

Felix Heide

18,235 次观看 • 8 个月前

Solar looks cheap on paper, but only because of two sneaky accounting tricks. First, the backup system is completely ignored: such as storage, balancing, and the dispatchable power plants waiting behind it. Second, fossil fuel prices are inflated via carbon taxes. In Germany, for example, domestic lignite (a low-grade brown coal) costs just 40 euros per MWh. But after carbon taxes, the cost triples to 120 MWh. Energy economist Lars Schernikau puts German solar at a whopping 10 times the full system cost of domestic lignite - once both the backup and carbon taxes are considered. Renewables are cheap on paper, but in reality are the most expensive source of power.
0:39

Solar looks cheap on paper, but only because of two sneaky accounting tricks. First, the backup system is completely ignored: such as storage, balancing, and the dispatchable power plants waiting behind it. Second, fossil fuel prices are inflated via carbon taxes. In Germany, for example, domestic lignite (a low-grade brown coal) costs just 40 euros per MWh. But after carbon taxes, the cost triples to 120 MWh. Energy economist Lars Schernikau puts German solar at a whopping 10 times the full system cost of domestic lignite - once both the backup and carbon taxes are considered. Renewables are cheap on paper, but in reality are the most expensive source of power.

Electroverse

29,356 次观看 • 4 个月前

We are thrilled to share that our first paper from my new lab, Spateo ( for spatiotemporal modeling of molecular holograms, is now online in Cell: Spateo is a comprehensive analytical framework for 3D whole-embryo spatiotemporal modeling. Its advanced features include: • 3D alignment and reconstruction at the whole-mouse-embryo scale (see the animation). • 3D spatial domain digitization and cell-cell communication analysis to understand spatial gene expression gradients and both inter- and intracellular communication. • 3D morphometric and volumetric analyses along with 3D morphogenesis vector field modeling to quantify dynamics such as surface area, volume, and cell density across organs, and to dissect the interplay between morphogenesis factors and cell migration. • A “Google Earth”-like browser, Spateo-viewer ( and for interactive and intuitive exploration of 3D spatial data. • Additional features, such as RNA signal-based single-cell segmentation. We are also honored that Nature “News and Views” has highlighted this work as well: This is really an amazing outcome after two years' heroic revision process that rewrite the entire paper using a new data ( for whole mouse embryos.
1:24

We are thrilled to share that our first paper from my new lab, Spateo ( for spatiotemporal modeling of molecular holograms, is now online in Cell: Spateo is a comprehensive analytical framework for 3D whole-embryo spatiotemporal modeling. Its advanced features include: • 3D alignment and reconstruction at the whole-mouse-embryo scale (see the animation). • 3D spatial domain digitization and cell-cell communication analysis to understand spatial gene expression gradients and both inter- and intracellular communication. • 3D morphometric and volumetric analyses along with 3D morphogenesis vector field modeling to quantify dynamics such as surface area, volume, and cell density across organs, and to dissect the interplay between morphogenesis factors and cell migration. • A “Google Earth”-like browser, Spateo-viewer ( and for interactive and intuitive exploration of 3D spatial data. • Additional features, such as RNA signal-based single-cell segmentation. We are also honored that Nature “News and Views” has highlighted this work as well: This is really an amazing outcome after two years' heroic revision process that rewrite the entire paper using a new data ( for whole mouse embryos.

evo-devo

58,508 次观看 • 1 年前

In a single generation, the UK has gone from having two offshore wind turbines to having just under 3,000. The industry now has a total generating capacity of 16.1 gigawatts (GW), which is enough to power more than 16 million homes each year, and over 600 million fully decorated Christmas trees 🎄 But where did it all start 25 years ago? 🤔 The Department for Energy Security and Net Zero Minister for Energy Michael Shanks MP is in Blyth to find out! 🔎 “Twenty-five years after the first offshore wind turbines began to turn, Britain is once again leading the world in clean homegrown power. “Offshore wind is at the heart of our 2030 mission – helping us reduce our dependence on volatile fossil fuel markets, lower bills for good, and support 100,000 jobs by 2030.” Find out more 👇 #Generational
1:03

In a single generation, the UK has gone from having two offshore wind turbines to having just under 3,000. The industry now has a total generating capacity of 16.1 gigawatts (GW), which is enough to power more than 16 million homes each year, and over 600 million fully decorated Christmas trees 🎄 But where did it all start 25 years ago? 🤔 The Department for Energy Security and Net Zero Minister for Energy Michael Shanks MP is in Blyth to find out! 🔎 “Twenty-five years after the first offshore wind turbines began to turn, Britain is once again leading the world in clean homegrown power. “Offshore wind is at the heart of our 2030 mission – helping us reduce our dependence on volatile fossil fuel markets, lower bills for good, and support 100,000 jobs by 2030.” Find out more 👇 #Generational

RenewableUK

31,362 次观看 • 9 个月前

Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors paper page: present Magic123, a two-stage coarse-to-fine approach for high-quality, textured 3D meshes generation from a single unposed image in the wild using both2D and 3D priors. In the first stage, we optimize a neural radiance field to produce a coarse geometry. In the second stage, we adopt a memory-efficient differentiable mesh representation to yield a high-resolution mesh with a visually appealing texture. In both stages, the 3D content is learned through reference view supervision and novel views guided by a combination of 2D and 3D diffusion priors. We introduce a single trade-off parameter between the 2D and 3D priors to control exploration (more imaginative) and exploitation (more precise) of the generated geometry. Additionally, we employ textual inversion and monocular depth regularization to encourage consistent appearances across views and to prevent degenerate solutions, respectively. Magic123 demonstrates a significant improvement over previous image-to-3D techniques, as validated through extensive experiments on synthetic benchmarks and diverse real-world images.
0:34

Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors paper page: present Magic123, a two-stage coarse-to-fine approach for high-quality, textured 3D meshes generation from a single unposed image in the wild using both2D and 3D priors. In the first stage, we optimize a neural radiance field to produce a coarse geometry. In the second stage, we adopt a memory-efficient differentiable mesh representation to yield a high-resolution mesh with a visually appealing texture. In both stages, the 3D content is learned through reference view supervision and novel views guided by a combination of 2D and 3D diffusion priors. We introduce a single trade-off parameter between the 2D and 3D priors to control exploration (more imaginative) and exploitation (more precise) of the generated geometry. Additionally, we employ textual inversion and monocular depth regularization to encourage consistent appearances across views and to prevent degenerate solutions, respectively. Magic123 demonstrates a significant improvement over previous image-to-3D techniques, as validated through extensive experiments on synthetic benchmarks and diverse real-world images.

AK

305,729 次观看 • 3 年前

After 15 years building biometric systems, Yanez.ai’s founder now has Bittensor miners trying to break them. On the latest Novelty Search, SN54 team explained how miners attack Yanez’s liveness and deepfake-detection systems with synthetic faces and screen replays, exposing weaknesses the team can fix and test again. The bigger goal is to prove that a user is real and unique without their biometric data ever leaving the device. Yanez says two clients are already fully integrated, with use cases ranging from Sybil-resistant Web3 participation to verifiable human authorization for payments. Hosted by const Full episode in the first comment
0:39

After 15 years building biometric systems, Yanez.ai’s founder now has Bittensor miners trying to break them. On the latest Novelty Search, SN54 team explained how miners attack Yanez’s liveness and deepfake-detection systems with synthetic faces and screen replays, exposing weaknesses the team can fix and test again. The bigger goal is to prove that a user is real and unique without their biometric data ever leaving the device. Yanez says two clients are already fully integrated, with use cases ranging from Sybil-resistant Web3 participation to verifiable human authorization for payments. Hosted by const Full episode in the first comment

Openτensor Foundaτion

25,876 次观看 • 2 个月前