IARPA Janus Benchmark Dataset
Technical information:
| Full name | |
|---|---|
| Country | |
| Contents | Facial Images |
| Images | 138,000 |
| Individuals | 3,531 |
| Runs database software | |
| URL"URL" is a type and predefined property provided by Semantic MediaWiki to represent URI/URL values. | |
| Related Technology |
Developers and Users:
| Developed by | IARPA |
|---|---|
| Owning institution | |
| Custodian institution |
Description[edit | ]
After the 2014 breakthrough of the DeepFace model’s human-level performance on facial recognition, there was a shift made to commercialize the technology. In 2015, NIST launched the IARPA Janus Benchmark-A face challenge (IJB-A) , which was an open challenge in which researchers executed algorithms on NIST-provided image sets, and returned output data to NIST for scoring. The competition was organized by the Intelligence Advanced Research Projects Activity (IARPA), an organization within the Office of the Director of National Intelligence. This challenge and its variations and updates IJB-B and IJB-C ran from 2015 to the end of 2017, growing into a dataset of 138,000 face images, 11,000 face videos, and 10,000 non-face images of celebrities and Internet personalities collected from the web. The 3,531 subjects included in the dataset are specifically designed not to overlap with the popular face recognition benchmarks at the time, such as VGG-Face and CASIA WebFace dataset, in order to prevent overfitting. 1
References
- ^ Raji, Inioluwa Deborah and Fried, Genevieve. About Face: A Survey of Facial Recognition Evaluation. , 2021.