Groups Similar Look up By Text Browse About

Similar articles
Article Id Title Prob Score Similar Compare
218171 ZDNET 2021-10-14:
Facebook: Here comes the AI of the Metaverse
1.000 Find similar Compare side-by-side
218122 VENTUREBEAT 2021-10-14:
Facebook introduces dataset and benchmarks to make AI more ‘egocentric’
0.979 0.635 Find similar Compare side-by-side
218463 ZDNET 2021-10-18:
Facebook: We want 10,000 new recruits to help build our VR metaverse
0.459 Find similar Compare side-by-side
217957 VENTUREBEAT 2021-10-11:
Facebook quietly acquires synthetic data startup AI.Reverie
0.370 Find similar Compare side-by-side
218247 ARSTECHNICA 2021-10-18:
Facebook AI moderator confused videos of mass shootings and car washes
0.353 Find similar Compare side-by-side
217943 ZDNET 2021-10-12:
Canva launches video suite as workers report significant increase in usage
0.350 Find similar Compare side-by-side
218187 ZDNET 2021-10-13:
Irish regulators support Facebook's 'consent bypass' legal maneuver, suggest $42 million fine for GDPR violations
0.335 Find similar Compare side-by-side
218121 THEVERGE 2021-10-13:
Facebook execs tease VR prototype hardware with new photos
0.327 Find similar Compare side-by-side
218461 ZDNET 2021-10-18:
Facebook hits back at claims its AI has minimal success in fighting hate speech
0.327 Find similar Compare side-by-side
218003 ZDNET 2021-10-11:
NEC scores deal to build Facebook transatlantic half-petabit cable
0.314 Find similar Compare side-by-side
217775 TECHREPUBLIC 2021-10-12:
Metaverse cheat sheet: Everything you need to know
0.308 Find similar Compare side-by-side
217868 THEVERGE 2021-10-13:
Facebook says it will ban content that sexually harasses celebrities
0.303 Find similar Compare side-by-side
218011 ZDNET 2021-10-11:
Amazon AWS's AI team seeks the profound in the industrial
0.302 Find similar Compare side-by-side
217810 THEVERGE 2021-10-13:
Why look at reality when you can edit what you see in real time?
0.301 Find similar Compare side-by-side
218002 ZDNET 2021-10-11:
Researchers develop AI system to improve eye disease detection
0.258 Find similar Compare side-by-side
217875 VENTUREBEAT 2021-10-12:
DeepMind is developing one algorithm to rule them all
0.257 Find similar Compare side-by-side
217948 VENTUREBEAT 2021-10-11:
DeepMind proposes new benchmark to improve robots’ object-stacking abilities
0.250 Find similar Compare side-by-side
218385 VENTUREBEAT 2021-10-15:
AI Weekly: AI model training costs on the rise, highlighting need for new solutions
0.248 Find similar Compare side-by-side
217776 VENTUREBEAT 2021-10-12:
AI edge chip startup Hailo lands $136M
0.245 Find similar Compare side-by-side
218081 THEVERGE 2021-10-14:
Facebook starts rolling out end-to-end encrypted backups to WhatsApp
0.245 Find similar Compare side-by-side
217934 VENTUREBEAT 2021-10-11:
Microsoft and Nvidia team up to train one of the world’s largest language models
0.240 Find similar Compare side-by-side
218417 VENTUREBEAT 2021-10-18:
State of AI Report tracks transformers in critical infrastructure
0.233 Find similar Compare side-by-side
217849 THEVERGE 2021-10-13:
Acer’s new ConceptD 7 SpatialLabs Edition laptop uses eye-tracking cameras for a glasses-free 3D display
0.231 Find similar Compare side-by-side
217781 VENTUREBEAT 2021-10-12:
AmplifAI’s data-powered people enablement platform gets a $18.5M investment
0.231 Find similar Compare side-by-side
217802 ZDNET 2021-10-13:
Opendoor discusses the secret sauce: 'A deeper mechanism to the world'
0.221 Find similar Compare side-by-side


ID: 218171


Date: 2021-10-14

Facebook: Here comes the AI of the Metaverse

Facebook has gathered thousands of hours of first-person video in order to develop neural networks that operate more capably with data seen from a first-person point of view. To operate in augmented and virtual reality, Facebook believes artificial intelligence will need to develop an "egocentric perspective. To that end, the company on Thursday announced Ego4D, a data set of 2,792 hours of first-person video, and a set of benchmark tests for neural nets, designed to encourage the development of AI that is savvier about what it's like to move through virtual worlds from a first-person perspective.  The project is a collaboration between Facebook Reality Labs and scholars from 13 research institutions, including academic institutions and research labs. The details are laid out in a paper lead-authored by Facebook's Kristen Grauman, " Ego4D: Around the World in 2.8K Hours of Egocentric Video." Grauman is a scientist with the company's Facebook AI Research unit. Her background as a professor at UT Austin has been focused on computer vision and machine learning in related topics.  The idea is that the data set will propel researchers to develop neural nets that excel at performing tasks from a first-person perspective -- in the same way that big datasets such as ImageNet propelled existing AI programs from a "spectator" perspective. The point of egocentric perception is to try to fix the problems a neural network has with basic tasks, such as image recognition when the point of view of an image shifts from third-person to first-person, said Facebook.  Also: Facebook announces $50 million investment in 'responsible' metaverse development Most image recognition systems that can detect objects seen from the sidelines have high failure rates if the object is presented the from the point of view of a person encountering the object. The Ego4D initiative specifically targets the Metaverse, the coming world of immersive social networking that Facebook CEO Mark Zuckerberg discussed at the company's last earnings report.  "These benchmarks will catalyze research on the building blocks necessary to develop smarter AI assistants that can understand and interact not just in the real world but also in the metaverse, where physical reality, AR, and VR all come together in a single space," said Facebook. The 2,792 hours of video were collected by Facebook staff using a variety of cameras. The Vuzix Blade augmented reality headset made by Vuzix is just one, others include GoPro, Pupil Labs, ZShades, and Wee-view. The purpose of mixing different sets is to avoid "over-fitting," write Grauman and collaborators, the phenomenon when a neural network has memorized frames of video information, rather than being tuned to infer similarities across differences. Facebook said the video was "captured by 750 unique camera wearers from 73 worldwide locations and 9 different countries. " Some of that was by Facebook staffers on the company's campus and some by the university collaborators. Also: Facebook brings metaverse to work with Horizon Workrooms (and you thought Zoom fatigue was bad) The "4D" in Ego4D references the temporal aspect of the video Facebook's staff spent 250,000 hours looking at and providing spoken narrations summarizing what's going on in the videos, with time-stamps attached.  Facebook says the narrations "are temporally dense," given that, "On average we received 13.2 sentences per minute of video, for a total of 3.85M sentences. In total the narrations describe the Ego4D video using 1,772 unique verbs (activities) and 4,336 unique nouns (objects)." The dataset is meant to be used to develop neural nets that will perform on a variety of new benchmark tests. To that end, Grauman and collaborators describe several new tests they've created that require a neural net to be able to produce a response to: tasks in the past, such as recall; tasks in the present, such as categorizing an activity; or future forecasting, such as producing a description of the result of an action.  For example, one task for a neural net could be to answer a natural-language query that requires the program to match the content of the query to a frame of video. An example is to ask the computer, "When did I read to my children?" The computer would have to find the scene where the camera wearer was reading to their kids. The task is labeled by the human annotations staff, who are given a pre-formatted list of labels and have to assign those to clips.  Facebook said they have 74,000 queries assigned in this way to 800 hours of video.  In a future prediction test, the computer might have to predict with which object in a frame of video the camera wearer will next interact. So, if they are at a table rolling dough, the next action predicted might be to grab a ball of dough on the table. The program will make the prediction by selecting one of a pre-set list of verbs that have been attached to video frames by the annotation staff, and appending a time estimation, like spitting out "take dough in 0.8 seconds." Also: Facebook already has your memories, smart glasses will get it moreThe data sets for Ego4D will be made available on Github next month, said Facebook. Users will be required to sign a data use agreement.