Actor Recognition

Actor Recognition is a face detection and recognition software based on Fraunhofer IDMT's Facial Analysis Technologies. It allows identification of individuals (actors, celebrities, moderators, or athletes, for example) in videos or photos. Using Actor Recognition large media and broadcasting archives can be searched through efficiently to quickly retrieve photo or video segments containing certain individuals.

Reduced time and effort

Actor Recognition facilitates the creation of metadata for large photo and video archives. Following a preparation process in the course of which the software is trained with imagery showing the face of an individual, it automatically searches for that person in photo and video archives, regardless how large the archive. Tedious manual search and evaluation of imagery is no longer required. As the analysis is content based, the preparation of the analysis can also be done without manually arranging, sifting, or indexing the content.

High recognition rate

Actor Recognition combines modules for automatic face detection and for face recognition. In addition, modules for visual quality analysis recognize and factor in the respective pose of the subject as well as fuzziness or deficits in the exposure of the photo or video. This allows the software to automatically identify the frames which are most appropriate for face recognition. If requested, the software can be trained to accommodate individual data and requirements.

Detailed results

Actor Recognition does not only recognize individuals in photos or videos, it also determines the exact position and length of segments identified within larger sequences.

Application Scenarios

  • Generation of metadata: Automatic creation of metadata tags in large photo and video archives
  • Searching for individuals in photo and video archives: Compilation of photos and video recordings containing certain individuals


  • Content-based analysis (no need for manual metadata generation)
  • Separate tools for training and application
  • User-defined thresholds for rejection of unknown individuals
  • Results stored in XML or SQL, other formats on request
  • Available as 32 and 64bit C++ Libraries for Windows, Linux , Mac OS X

Learn more

Business Unit Media Management and Delivery

Facial Analysis Technologies