Detecting and Tracking Regions of Interest for Remote Measurement of Vital Parameters

Sammanfattning

The initial assessment of a mass casualty incident is essential to effectively conducta rescue operation. The survival rate is affected by the complexity of the incident,and it is therefore imperative to enhance the operational capacities of emergencymedical services and civil protection agencies in mass casualty incidents. This thesisinvestigates the possibilities for an unmanned aerial vehicle (UAV) to detect andtrack regions of interest for remote measurement of vital parameters in visual andthermal footage for first response triage purposes. The regions of interest are thenose, mouth, and chest, and the UAV characteristic taken under consideration inthis thesis is image blur due to random camera motion. In this thesis, we take an objectdetection approach and implement the keypoint estimation framework KAPAOand the tracking algorithm SORT in several different experimental setups. UsingKAPAO and SORT, we achieve a good result. For the detection in the thermaldomain, the model created by transferring knowledge from the visual to the thermaldomain achieves the highest performance. We also consider adversarial trainingon random motion blur, however the result shows a minimal impact on the modelperformance in the presence of characteristic low-altitude UAV motion blur. Regardingthe tracking of the regions of interest, the result concludes that the SORTalgorithm improves the performance compared to assigning tracking identificationnumbers based on frame-to-frame differences. The result shows that the distanceto the subjects and the image quality impacts the performance. Compared withprevious work on remote measurement of vital parameters, the algorithms of thisthesis achieve a nearly perfect score on corresponding distances. If the distancesare realizable in a UAV triage application is however unknown and has to be investigatedfurther. Moreover, the work of this thesis problematizes the low-altitudeUAV motion blur which poses a potential limitation in a potential UAV triage application.An alternative could hence be to use optical stabilization measurementfor blur reduction.