Deep Learning Detection of Artificial Light Effects on Nocturnal Mammal Behaviour

Abstract

Introduction: Light pollution from artificial sources has become an important issue in ecology, which has greatly influenced the normal behaviour and survival techniques of nocturnal mammals. With the increase in urbanization and lighting of roads as well as artificial light at night, the natural environment of darkness has been disturbed due to which movement, feeding habits, reproductive cycles, predator-prey relationships, and habitat usage of wild animals has been disrupted. As nocturnal mammals rely on the natural darkness for various biological functions, light pollution may affect their behaviour and prevent them from adapting to the environment. Traditional wildlife observation techniques are difficult and time-consuming. The proposed research project discusses the development of a deep learning framework for automatic analysis of the impact of artificial light on nocturnal behavior of mammals through computer vision techniques. The proposed framework includes the use of deep learning models for identification of mammals' species, analysis of their behavior, and analysis of differences in their activity depending on lighting conditions. A computer vision pipeline will be designed to process images and videos of night wildlife to perform automatic detection of behavioral response to the exposure of light, such as activity change and habitat preferences. The experimental research shows that the proposed deep learning technique is effective in recognizing nocturnal mammal activities and detecting variations in their behavior related to the exposure to artificial light. It proves the potential of artificial intelligence systems for monitoring the wildlife and conducting environmental research