Depression Detection From Eye Blink Features

Abstract

Depression is a widespread mental health disorder which burdens over 350 million people world-wide. Current depression screening and assessment methods rely almost completely on clinical interviews and self-report scales. While being beneficial, such measures need objective and effective ways of integrating behavioural observations that are strong symptoms of depression presence and severity. A system with the potential of serving as a decision support system is proposed based on eye blink features extracted using a racial landmark tracker. The performance of eye blink features extracted from video frames has been analysed for a binary classification task (depressed vs. non-depressed). The proposed system evaluated using several classification schemes on the dataset provided by the Audio/Visual Emotion Challenge of 2013 and 2014. We find that eye blink features gave 92.95% accuracy using adabost classifier for north- wind task and 88% for freeform task over the entire overview. Classification accuracy was higher for the text reading task than for the answering questions task for all test scenarios. These findings suggest that automatic classification of depression from eye blink features in patients is achievable.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 36 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.