Jump to content

DeepFace

From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by 176.40.35.139 (talk) at 21:30, 12 December 2020 (comparison with google facenet). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

DeepFace is a deep learning facial recognition system created by a research group at Facebook. It identifies human faces in digital images. It employs a nine-layer neural network with over 120 million connection weights and was trained on four million images uploaded by Facebook users.[1][2] DeepFace shows human-level performance. The Facebook Research team has stated that the DeepFace method reaches an accuracy of 97.35% ± 0.25% on Labeled Faces in the Wild (LFW) data set where human beings have 97.53%.[3] This means that DeepFace is sometimes more successful than the human beings. However, DeepFace model falls behind Google FaceNet which got 99.65% on the same data set.[4] It also leaves behind the FBI's Next Generation Identification system which have 85% performance [5] One of the creators of the software, Yaniv Taigman, came to Facebook via their 2007 acquisition of Face.com.[6]

Commercial rollout

Facebook started rolling out the technology to its users in early 2015, with the exception of users in the EU due to data privacy laws there.[citation needed]

Academic analysis

The software was the subject of graduate-level artificial intelligence (AI) coursework in 2015.[7]

Reactions

AI researcher Ben Goertzel said Facebook had "pretty convincingly solved face recognition" with the project, but said it would be incorrect to conclude that deep learning is the entire solution to AI.[8]

A Huffington Post piece called the technology "creepy" and, citing data privacy concerns, noted that some European governments had already required Facebook to delete facial-recognition data.[9] According to Broadcasting & Cable, both Facebook and Google had been invited by the Center for Digital Democracy to attend a 2014 National Telecommunications and Information Administration "stakeholder meeting" to help develop a consumer privacy Bill of Rights, but they both declined.[10] Broadcasting & Cable also noted that Facebook had not released any press announcements concerning DeepFace, although their research paper had been published earlier in the month.[10] Slate said the lack of publicity from Facebook was "probably because it's wary of another round of 'creepy' headlines".[11]

See also

References

  1. ^ "Facebook creates software that matches faces almost as well as you do", Technology Review, Massachusetts Institute of Technology, March 17, 2014
  2. ^ Facebook's DeepFace shows serious facial recognition skills, CBS News, March 19, 2014
  3. ^ "DeepFace: Closing the Gap to Human-Level Performance in Face Verification". Facebook Research. Retrieved 2019-07-25.
  4. ^ "LightFace: A Lightweight Deep Face Recognition Framework". IEEE. Retrieved 2020-12-13.
  5. ^ Russell Brandom (July 7, 2014), "Why Facebook is beating the FBI at facial recognition", The Verge
  6. ^ Amit Chowdhry (March 18, 2014), "Facebook's DeepFace Software Can Match Faces With 97.25% Accuracy", Forbes
  7. ^ Richard Baraniuk (2015), ELEC 631 syllabus, Rice University
  8. ^ Ben Goertzel (March 22, 2014), "Lessons from Deep Mind & Vicarious", The Multiverse According to Ben (blog)
  9. ^ Dino Grandoni (March 18, 2014), "Facebook's New 'DeepFace' Program Is Just As Creepy As It Sounds", The Huffington Post
  10. ^ a b John Eggerton (March 24, 2014), CDD: Google, Facebook Decline to Present at Facial Recognition Meeting: Had been asked by nongovernmental groups for more info on technology
  11. ^ Will Oremus, "Facebook's New Face-Recognition Software Is Scary Good", Slate

Further reading