Detection and Identification of Chimpanzee Faces in the Wild

In this paper, we present and evaluate a unified automatic image-based face detection and identification framework using two datasets of captive and free-living chimpanzee individuals gathered in uncontrolled environments. This application scenario implicates several challenging problems like different lighting situations, various expressions, partial occlusion, and non-cooperative subjects. After the faces and facial feature points are detected, we use a projective transformation to align the face images. All faces are then identified using an appearance-based face recognition approach in combination with additional information from local regions of the apes' face. We conducted open-set identification experiments for both datasets. Even though, the datasets are very challenging, the system achieved promising results and therefore has the potential to open up new ways in effective biodiversity conservation management.