FaceEdutain – Computer based identification of great apes

FaceEdutain is a system for automatic identification of great apes in zoos and wildlife parks. It uses intelligent face recognition software developed by Fraunhofer IIS and IDMT to identify individual animals by means of distinctive facial features, such as individual patterns of wrinkles under the ape’s eyes, or specific colorings. FaceEdutain requires only a short period of training (which means that the system simply needs to be fed with pictures of each animal to be identified) to present visitors information about each individual ape.

Interactive learning about apes in zoos and wildlife parks

Instead of reading only static general information about ape populations given on boards positioned next to the compound, zoo visitors may now request information on each individual animal by directing a camera towards selected apes, or small groups of apes, and have them automatically identified by the system, in order to learn the ape’s name, age, origin, kinship with other apes around, and so on.

Edutainment for visitors

Provided the system has been fed with proper information, visitors may also learn about the preferences and character traits of individual animals, and about the social relationships existing in certain groups. As the system offers also background knowledge on the behavior and social interaction of the apes, visitors can submerge themselves into the world of the apes they are watching.

• System works reliably even under varying ambient lighting conditions, or when the apes’ faces are partially concealed
• System requires only little training (simply needs to be fed with 20 to 30 pictures of each ape)
• High face recognition rate even when apes do not look straight into the camera (as long as system is able to detect both eyes of the ape)
• System can easily be fed with data of new animals
• Real-time capable

The scientific basis of FaceEdutain was laid in an collaborative R&D project named SAISBECO (»Semi-Automated Audiovisual Species and Individual Identification System for Behavioral Ecological Research and Conservation«), featuring participation of Max Planck Institute for Evolutionary Anthropology, the University of Bristol, Fraunhofer IIS, and Fraunhofer IDMT.

The aim of the three-year project was to relieve researchers and park rangers from the tedious and time-consuming work of looking through tons of images to determine the size of great ape populations in the wilderness. The software developed in the course of the project scans the recorded material for sequences showing the faces of individual apes, who are then identified by their distinctive facial features. The system does this at a fraction of the time needed for manual evaluation of material. And it even allows parallel detection and identification of several individuals within a group.

Alongside with video data, the software also evaluates audio material recorded. Using the audio analysis tools developed by Fraunhofer IDMT, the software is able to distinguish different sounds and cries, and assign them to individual apes.