University of Twente Student Theses
Investigating the feasibility of using a RealSense depth camera D435i by creating a framework for 3D pose analysis
Shahmoradi, Reihaneh (2022) Investigating the feasibility of using a RealSense depth camera D435i by creating a framework for 3D pose analysis.
|
PDF
8MB |
| Abstract: | The main task in the present study is to investigate the feasibility of using an optimum depth camera to monitor step locations. The Real Sense D435i depth camera is selected for its user friendliness, low cost and ease of availability. In the general framework used, RGB results from RealSense camera are fed into a software, which is based on machine learning algorithm. This software, called OpenPose, produces the required joint data on human anatomy. Back projection for 3D pose estimation is then done by using the depth values and the output of OpenPose. Filtering of the data is the last stage in the applied framework. The accuracy of step locations is found by comparing the results with golden standard (Qualisys) measurements. The highest accuracies are obtained for forward steps (steps towards the camera). The results show errors of 1.1 cm and 1.12 cm for right and left forward steps at a distance of 293 cm. The signal to noise ratio (SNR) values are derived and again the mentioned forward steps show better results i.e. 19.73 and 18.9 respectively. The higher distance to the camera is assessed to be the main cause of lower accuracy in the results of backward steps. The occlusion occurs in some images. The occlusion effect in these images appears as a fake peak in the time series graph for forward steps. This effect appears as a missing the peak for backward steps. The effected results have been left out in the calculations of accuracy assessment. Further works could be performed by using more depth cameras of this type to gain more efficient measurements in all directions. Keywords: exergames, RealSense D435i, Qualisys, stepping accuracy, SNR, occlusion |
| Item Type: | Essay (Master) |
| Clients: | Unknown organization, Enschede |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Programme: | Electrical Engineering MSc (60353) |
| Link to this item: | https://purl.utwente.nl/essays/92232 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page