Repository logo

Browsing by Author "Schofield, Daniel"

Now showing 1 - 2 of 2
  • Automated audiovisual behavior recognition in wild primates
    Publication . Bain, Max; Nagrani, Arsha; Schofield, Daniel; Berdugo, Sophie; Bessa, Joana; Owen, Jake; Hockings, Kimberley J.; Matsuzawa, Tetsuro; Hayashi, Misato; Biro, Dora; Carvalho, Susana; Zisserman, Andrew
    The field of ethology seeks to understand animal behavior from both mechanistic and functional perspectives and to identify the various genetic, developmental, ecological, and social drivers of behavioral variation in the wild (1). It is increasingly becoming a data-rich science: Technological advances in data collection, including biologgers, camera traps, and audio recorders, now allow us to capture animal behavior in an unprecedented level of detail (2). In particular, large data archives including both audio and visual information have immense potential to measure individual- and population-level variation as well as ontogenetic and cultural changes in behavior that may span large temporal and spatial scales. However, this potential often goes untapped: The training and human effort required to process large volumes of video data continue to limit the scale and depth at which behavior can be analyzed. Automating the measurement of behavior can transform ethological research, open up large-scale video archives for detailed interrogation, and be a powerful tool to monitor and protect threatened species in the wild. With rapid advances in deep learning, the novel field of computational ethology is quickly emerging at the intersection of computer science, engineering, and biology, using computer vision algorithms to process large volumes of data (3).
    2021-11Journal article Open access Show more
  • Chimpanzee face recognition from videos in the wild using deep learning
    Publication . Schofield, Daniel; Nagrani, Arsha; Zisserman, Andrew; Hayashi, Misato; Matsuzawa, Tetsuro; Biro, Dora; Carvalho, Susana
    Video recording is now ubiquitous in the study of animal behavior, but its analysis on a large scale is prohibited by the time and resources needed to manually process large volumes of data. We present a deep convolutional neural network (CNN) approach that provides a fully automated pipeline for face detection, tracking, and recognition of wild chimpanzees from long-term video records. In a 14-year dataset yielding 10 million face images from 23 individuals over 50 hours of footage, we obtained an overall accuracy of 92.5% for identity recognition and 96.2% for sex recognition. Using the identified faces, we generated co-occurrence matrices to trace changes in the social network structure of an aging population. The tools we developed enable easy processing and annotation of video datasets, including those from other species. Such automated analysis unveils the future potential of large-scale longitudinal video archives to address fundamental questions in behavior and conservation.
    2019Journal article Open access Show more