Automatic Monitoring of Rat Behavior: Leveraging Deep Neural Networks for Accurate Classification

Abstract

The analysis of animal behavior is crucial in fields such as medicine, biomedical research, and neuroscience, as it provides insights into both physiological and psychological aspects of various species. Conventionally, this requires human observation, which is labor-intensive, time-consuming, and prone to errors. Recently, convolutional neural networks have demonstrated remarkable success in image and video processing across diverse applications. In this study, we investigate the use of convolutional neural networks to analyze rat behavior in a controlled laboratory setting. Using a ResNet-18 deep neural network, we classify rat behaviors from images captured in Skinner boxes under varying experimental conditions. Our approach results in a near-perfect classification accuracy, highlighting the effectiveness of deep learning models for automated animal behavior analysis, offering a scalable and efficient alternative to direct observation.