JoWUA

Abstract

Protection of wildlife is one of the important strategies for maintaining the ecological balance of the Earth and for sustainable development of our biodiversity. Understanding the effects of such activities on animal behavior can be crucial to managing them to protect or rehabilitate local wildlife — but to do that, you need to know where, when, how, and how much animals are active. Conventional tracking is based on visual observation of animal movements, but it is a manual operation with limitations in observing certain motion patterns and behaviors. To address these problems, sensors integrated with deep learning have emerged as the 'save grace' for the wildlife conservation sector. Lastly, it is possible to attach sensors (such as GPS and accelerometers) to wildlife to capture their behaviors and demeanor in situ. The produced data is large and complex and may be tedious to analyze by hand. Such sensors can provide incredibly detailed and accurate data on animal behavior using deep learning algorithms, a type of computer program that learns and makes predictions from large datasets. Sensors and deep learning combined enable models to observe and forecast animal behavior with ease. We can use them to sense and classify various behaviors (such as feeding, mating, and migration) precisely in real time. They can discriminate different movement paths by their indication of changes in habitat quality or anthropogenic disturbance. Thanks to this, the data can help 'conservationists make the right decisions and strategies for saving animal populations.