Kullback-Leibler Divergence Based Method for Depression Diagnosis Using Video Data
Harsh Bhasin, Nishant Kumar, Anupama Singh, Sharma Manish, Ram Pratap Beniwal · 2023
Depression affects around four per cent of the total world population. The lack of trained professionals in low and middle-income countries has led to more than three-quarters of those affected receiving no treatment. This calls for the development of automated methods for depression detection. This work proposes a model to detect pertinent frames from video data using Kullback-Leibler Divergence and the concept of outliers. The distribution of microstructures of the frames, selected using the proposed method, is fed to the Support Vector Machine to classify those affected with depression from the controls. The proposed method is validated using the data obtained from the Department of Psychiatry, Center of Excellence in Mental Health, Atal Bihari Vajpayee Institute of Medical Sciences and Dr Ram Manohar Lohia Hospital, New Delhi. The results are encouraging and pave the way for the use of this method in developing systems for the automated detection of the disease.