Deep Learning Analysis of Exploratory Behaviour in Environmentally Enriched Mammals

Abstract

Exploratory enrichment is a vital strategy towards improving the wellbeing of mammals through stimulation of their natural behaviors such as exploring, moving around, object manipulation and environmental inspection. Understanding the behavior of animals in captivity and managed environments is key in evaluating their psychological state of mind, adaptability and quality of habitat. Unfortunately, the existing methods of behavioral analysis largely rely on manual observation and annotation and are labor-intensive, non-scalable and are associated with observation bias. To address the above limitations, this work introduces a deep learning-based framework for behavioral monitoring in exploratory enriched mammals. The framework relies on animal video data sets under enriched environments and employs deep learning models in behavioral detection and classification. The model learns to detect complex visual cues related to exploratory behaviors, making it possible to detect various behavioral states and motions. It was demonstrated through experiments that the suggested method makes it possible to detect behavioral differences and automate the process of monitoring more effectively than traditional observation. The suggested framework allows for the continuous monitoring of behavioral patterns without involving many manual actions, thus providing more efficient analysis. Moreover, the suggested method is helpful for assessing the enrichment efforts, improving animal welfare, and conducting behavioral ecology studies. The combination of deep learning and animal monitoring tools can be a breakthrough in the field of wildlife intelligent observation and conservation planning