Fraunhofer Center for Assistive Information and Communication Solutions – AICOSDescription:
This thesis proposes the development of an algorithm for the automatic labelling of markers in MoCap data. The tool, to be developed in Python, should combine information about markers’ dynamics and intermarkers relationships to provide reliable results. The algorithm should be robust to the presence of missing marker information and be able to operate in distinct conditions (e.g., different marker sets, different movements, etc.). It should also report the data frames where labelling uncertainty was higher, so that, whenever needed, manual correction can be better targeted and faster. The performance of the algorithm shall be reported using previously recorded (and manually post-processed) datasets (e.g. the CMU Graphics Lab Motion Capture Database, or the Motion Capture Database HDM05).
Outcome:
This thesis intends to continue the work already started at FhP-AICOS, concerning marker labelling and gap filling for MoCap systems. With this work, we aim to value and promote the continued use of VICON in our lab (e.g., facilitate the use by our collaborators; offer external services; etc.). The results of the thesis may also bring additional value to the MoCap industry.
Author: José Gomes
Type: MSc thesis
Partner: FEUP – Faculdade de Engenharia da Universidade do Porto
Year: 2022