Fig. 2. Behavior recognition results demonstrate the CNN model’s robustness to variations in pose, lighting, scale, and speed of action.

Example of correctly labeled body tracks from unseen and unheard videos (nut cracking and drumming for the top two and bottom two rows, respectively). Middle two rows: Multiple individuals nut-cracking and buttress-drumming showing variations in lighting, pose, background, and number of chimpanzees. Photo credit: Kyoto University, Primate Research Institute; The Cantanhez Chimpanzee Project.