• Alstrup, Aage Kristian Olsen
• Ishøj, Matilde
• Overgaard, Clara
• Pertoldi, Cino
• Thaarup, Sarah Risager
• Jensen, Trine Hammer
• Andersen, Sebastian Vadskær
• Bakke, Inge Kathrine

Abstract

Machine learning techniques have been used for observing zoo animals and quantifying their behaviour. This study investigates the behaviour of black lemurs (Eulemur macaco) and ring-tailed lemurs (Lemur catta) in a new walk through enclosure at Aalborg Zoo in Denmark, on a day with many visitors (1,031 visitors) and on a day with fewvisitors (181) for observing possible differences in four types of lemur behaviors: ‘locomotion’, ‘resting’, ‘eating’,and ‘grooming’. By using both manually observed data and machine learning, this study compares the methods and explores the lemurs´ behaviours. The Wilcoxon rank sum tests of the four behaviours manually estimated in the two days for ring-tailed lemurs showed that the visitors were significantly affecting several of these behaviours. However, locomotor activity was not found to be significantly different in the two days for black lemurs, but when testing the data obtained with the machine learning approach, a significant difference between days was found. The results suggest that the manual approach can complement a machine learning approach in behavioural studies.

Topics

• machine learning