Sensei: A Multi-Modal Framework for Assessing Stress Resiliency
Abstract
Progress during this reporting period was in two primary areas. First, we completed all necessary modifications to our laboratory environment to allow accurate, automatic data collection in Experiment 2. This included: (1) Enabling full time stamping of all data streams (IR and visible video sensors, plus physiological signals from the Equivital belt). (2) Developing automatic, trial synced audio recording. (3) Better illumination to allow pupillometry from the high resolution video camera. (4) Time-stamping of subject responses to the peripheral detection task. (5) Python-based automatic start/stop of all data streams. (6) Automatic generation and population of directories for each data stream for each subject. In addition to these lab accomplishments, we also fine-tuned the timing of the Stroop/PDT presentations to achieve maximal engagement and hence maximal stress response as the trials come more and more quickly towards the end of the session. In data analysis, we perfected a technique, described briefly in the last report, for accurately tracking facial feature points in the IR video stream. As shown in Figure 1, this tracking works quite well, even over large facial rotations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2013
- Accession Number
- ADA582457
Entities
People
- Ajay Divakaran
- Jeffrey Lubin
- Joe Ferraro
Organizations
- SRI International