The ability to identify individual animals is a prerequisite for many questions in behavioral ecology, cognitive research, conservation monitoring, and wildlife epidemiology. However, for a number of cryptic or elusive species it is currently impossible to efficiently identify large numbers of individuals and their associated traits, which is necessary for studying inter-population differences regarding social dynamics and demography. Individual identification is time-consuming in the wild, in particular for elusive species. In captive settings, this work can also be tedious, when large amounts of data (e.g. video) need to be processed. With the increasing availability of remote audiovisual recording devices, such as camera traps, standardized data collection has become much easier, in particular in the wild. Yet there are no techniques available to rapidly process these data, and consequently, there is a high demand for semi-automated approaches allowing the identification of species, age and sex classes, and in particular the identification of individuals from visual and acoustic recordings.
In the joint effort of the Fraunhofer Institutes IDMT and IIS, and MPI EVA, we have the following three objectives:
- extensive data collection using remote video cameras and autonomous audio devices in the field and in a captive environment.
- evaluation and development of several preprocessing techniques, such as multiple-object-tracking and face-detection, as well as the development of algorithms for the identification of species, age and sex classes, and the semi-automatic recognition of individual animals using different audio-visual features.
- subsequent analysis of the processed data to demonstrate the value of this approach for behavioral-ecological research and conservation, i.e. social contact network-analysis for wildlife epidemiology studies, rapid assessment of biodiversity data, demographic and population monitoring analysis, and behavioral patterns of captive individuals.
Great apes and other primates are the primary focus of our study, but we will include other species as well to demonstrate the wide applicability of the approaches taken.
This project will pioneer the use of an innovative sampling and processing approach for acquiring large amounts of data on species and individual animals. It will thus address the current demand in behavioral ecological research and conservation for increased sample sizes and standardized data quality. It thus has the potential to open up new venues in these fields, by reducing the burden of time-consuming routine work and thus releasing previously bound resources for innovative research projects.