Behavioral features fusion for ethological CNN classification of open field test videos

Zhaolin Xiao, Huan Liu, Guoqing Zhou, Feng Zhu, Haiyan Jin

Abstract: In both ethological and pharmacological experiments, open field test (OFT) is a classic experiment for measuring mouse general activity and exploratory behaviors. The OFT manual analysis is usually a time consuming and costly process. In this paper, we propose a convolutional neural network (CNN) based classification, which can automatically sort out two groups of mice with different behaviors. We first extract the multi-modality features from both spatial and time domain of the OFT video. Then, all extracted features are regularized and fused to a 2D behavior feature map. Finally, we train a CNN network to distinguish mice with different behaviors. On real OFT datasets experiments, the proposed classification significantly outperforms other methods.