Difference between revisions of "DeepFace (dataset)"
| Line 3: | Line 3: | ||
|Creation Date=2014 | |Creation Date=2014 | ||
|Dataset Category=Facial Recognition | |Dataset Category=Facial Recognition | ||
| + | |Related Technology=DeepFace | ||
}} | }} | ||
==Description== | ==Description== | ||
<blockquote>It was not until the breakthrough of Alexnet in 2012, and the subsequent introduction of the DeepFace model in 2014, that the use of neural networks became a mainstream method for facial recognition development. DeepFace, the first facial recognition model trained with deep learning, was also the first instance of a facial recognition model approaching human performance on a task. Deepface was developed by researchers at Facebook, Inc. and trained on an internal dataset composed of images from Facebook profile images; at the time, it was purportedly “the largest facial dataset to-date, an identity labeled dataset of [[Has images::4 million|4 million]] facial images belonging to more than [[Has individuals::4,000|4,000]] identities” (Taigman et al. 2014).The impact of deep learning techniques on face recognition and its adjacent problems was dramatic; the DeepFace model achieved a 97.35% accuracy on the Labeled Faces in the Wild (LfW) test set, reducing the previous state of the art’s error by 27%. [[CiteRef::rajiFaceSurveyFacial2021]]</blockquote> | <blockquote>It was not until the breakthrough of Alexnet in 2012, and the subsequent introduction of the DeepFace model in 2014, that the use of neural networks became a mainstream method for facial recognition development. DeepFace, the first facial recognition model trained with deep learning, was also the first instance of a facial recognition model approaching human performance on a task. Deepface was developed by researchers at Facebook, Inc. and trained on an internal dataset composed of images from Facebook profile images; at the time, it was purportedly “the largest facial dataset to-date, an identity labeled dataset of [[Has images::4 million|4 million]] facial images belonging to more than [[Has individuals::4,000|4,000]] identities” (Taigman et al. 2014).The impact of deep learning techniques on face recognition and its adjacent problems was dramatic; the DeepFace model achieved a 97.35% accuracy on the Labeled Faces in the Wild (LfW) test set, reducing the previous state of the art’s error by 27%. [[CiteRef::rajiFaceSurveyFacial2021]]</blockquote> | ||
Revision as of 19:49, 26 February 2021
Technical information:
| Full name | |
|---|---|
| Country | |
| Contents | Facial Images |
| Images | 4,000,000 |
| Individuals | 4,000 |
| Runs database software | |
| URL"URL" is a type and predefined property provided by Semantic MediaWiki to represent URI/URL values. | https://pypi.org/project/deepface/ |
| Related Technology | DeepFace |
Developers and Users:
| Developed by | |
|---|---|
| Owning institution | |
| Custodian institution |
Description[edit | ]
It was not until the breakthrough of Alexnet in 2012, and the subsequent introduction of the DeepFace model in 2014, that the use of neural networks became a mainstream method for facial recognition development. DeepFace, the first facial recognition model trained with deep learning, was also the first instance of a facial recognition model approaching human performance on a task. Deepface was developed by researchers at Facebook, Inc. and trained on an internal dataset composed of images from Facebook profile images; at the time, it was purportedly “the largest facial dataset to-date, an identity labeled dataset of 4 million"million" can not be assigned to a declared number type with value 4. facial images belonging to more than 4,000 identities” (Taigman et al. 2014).The impact of deep learning techniques on face recognition and its adjacent problems was dramatic; the DeepFace model achieved a 97.35% accuracy on the Labeled Faces in the Wild (LfW) test set, reducing the previous state of the art’s error by 27%. 1
References
- ^ Raji, Inioluwa Deborah and Fried, Genevieve. About Face: A Survey of Facial Recognition Evaluation. , 2021.