A Pharmacology Toolkit for Animal Pose Estimation, Tracking and Analysis

Pose estimations have increased dramatically in the realm of computer vision since the introduction of sensor camera technology and the development of deep networks. Since predicting animal poses is a critical first step in understanding animal behaviour, animal position studies are included in the research on pose estimations. Two different studies were used to identify the crossing, rearing and grooming behaviours in an open field experiment. The first concentrated on crossing while the second concentrated on grooming and rearing. After employing various picture processing techniques, both trials were able to count the behaviours. Through the use of computer vision, deep learning networks, euclidean distance matching algorithms, and image processing, The desired output of classifying and tracking several mouse behaviours like crossing and rearing in an open-field experiment are achieved. The rearing experiment resulted in an accuracy of 97% across all mice.