Evaluation of Face Recognition APIs and Libraries

dc.contributor.authorMasek, Philip
dc.contributor.authorThulin, Magnus
dc.contributor.departmentGöteborgs universitet/Institutionen för data- och informationsteknikswe
dc.contributor.departmentUniversity of Gothenburg/Department of Computer Science and Engineeringeng
dc.date.accessioned2015-05-06T08:17:26Z
dc.date.available2015-05-06T08:17:26Z
dc.date.issued2015-05-06
dc.description.abstractAfter years of research, the commercialization of face recognition technology is apparent with the emergence of several face recognition libraries and APIs. Organizations and developers are faced with identifying critical success factors when selecting a face-recognition API or library to be used within the development of a product. This study aims to (1) understand which quality characteristics derived from the ISO/IEC 25010 standard are important for an organization adopting the technology and (2) evaluate two client-side libraries and two cloud based APIs according to the quality characteristics identified. Data was extracted by interviewing a company investigating face recognition technology for software-reuse and an experiment was carried out to evaluate the chosen software by extracting metrics from the ISO 9126 standard. An organization adopting face recognition technology prioritised reliability, functional suitability and maintainability as the most important. The experiment concluded the chosen cloud-based APIs were more computationally accurate than the client-side libraries. However, the data collected concluded that the chosen client-side libraries have less failure density than cloudbased APIs.sv
dc.identifier.urihttp://hdl.handle.net/2077/38856
dc.language.isoengsv
dc.setspec.uppsokTechnology
dc.titleEvaluation of Face Recognition APIs and Librariessv
dc.typetext
dc.type.degreeStudent essay
dc.type.uppsokM2

Files

Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
gupea_2077_38856_1.pdf
Size:
952.84 KB
Format:
Adobe Portable Document Format
Description:
Bachelor Thesis

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
876 B
Format:
Item-specific license agreed upon to submission
Description: