Timely detection of depression and the accurate assessment of its severity are the two major challenges that face the medical community. To assist the clinicians, various objective measures are being explored by researchers. In literature, features extracted from the images or videos, are found relevant for detection of depression. Various feature extraction methods are suggested in literature. However, the high dimensionality of the features so obtained provide an overfitted learning model. This is handled in this work with the help of three popular univariate filter feature selection methods, which identify the reduced size of relevant subset of features. The combinations of univariate techniques with well-known classification and regression techniques are investigated. The performance of classification and regression techniques improved with the use of feature selection methods. Moreover, the proposed model has outperformed most of the video-based existing methods for identifying depression and determining its level of severity.
CITATION STYLE
Rathi, S., Kaur, B., & Agrawal, R. K. (2019). Enhanced Depression Detection from Facial Cues Using Univariate Feature Selection Techniques. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11941 LNCS, pp. 22–29). Springer. https://doi.org/10.1007/978-3-030-34869-4_3
Mendeley helps you to discover research relevant for your work.