Application of local binary patterns and cascade AdaBoost classifier for mice behavioural patterns detection and analysis

Abstract

The paper describes the application of local binary patterns and cascade AdaBoost classifier (CAC) to detect and analyse mice behavioural movement. This was done with a view to investigating the inconsistencies associated with current practices, whereby mice behavioural classification is achieved by means of human-generated labels. The developed cascade AdaBoost algorithm was able to detect eight different mice movement, and we develop a system that allows mice behavioural analysis in videos, with minimal supervision. Evaluating the results on Completeness, Consistency and Correctness, and based on the devised analysis, a solution was deployed, showing that machine learning plays an important role in translating video data into scientific knowledge. This is a useful addition to the animal behaviourist’s analytical toolkit.