Facial thermal and blood perfusion patterns of human emotions: Proof-of-Concept

Contact-Free Emotion Recognition for Monitoring of Well-Being: Early Prospects and Future Ideas

“…Last but not least, thermal imaging has also been investigated in emotion recognition. Aristizabal-Tique et al [11] gathered thermographic data on participants in three different conditions -baseline, positive, and negative valence of emotions. Analysing the data, blood perfusion and average temperature was calculated for the regions of interest (ROIs).…”

“…Thermal imaging can be used to detect changes in temperature between ROIs, which correspond to changes in psychological or emotional state. A recent example dataset containing such thermal recordings with corresponding emotion labels was provided by Aristizabal-Tique et al [11] and was described in Section 2. Our aim was to use their thermal data for precise ROI segmentation and verify the segmentation performance on a consumer thermal camera with lower specifications (resolution), which would be feasible to use in our planned larger-scale experiments.…”

“…As outlined in Section 2, thermal imaging was shown in literature to be an efficient approach to emotion recognition. Thus, we investigated the dataset provided by Aristizabal-Tique et al [11], described in the Section 3.2, which contains videos of four emotional states (baseline, neutral, fear, happiness) for each of the three subjects. The videos were shot in front of a computer screen, similar to the setting that we intend to utilize in our upcoming study.…”

“…The videos were shot in front of a computer screen, similar to the setting that we intend to utilize in our upcoming study. Initially, we decided to replicate the approach and findings of Aristizabal-Tique et al [11]. Furthermore, our aim was to test the employed methods on data coming from the FLIR Lepton thermal camera, which is more feasible for use in our upcoming larger-scale experiments due to its substantially lower price.…”

“…In the next step after extracting relevant facial landmarks, we used them to identify ROIs. The ROIs (nose, forehead, and eyes) were chosen based of their known connection to psychological changes, as reported in related work [11,12]. When comparing average temperatures of the ROIs for each of the four conditions, we discovered some changes across conditions and compared them to the work of Aristizabal-Tique et al [11].…”