Task event annotation using software ATLAS
Lígia Reis Nóbrega, Adriano de Oliveira Andrade, Adriano Alves Pereira · Zenodo (CERN European Organization for Nuclear Research) · 2022
Data annotation is the process of labeling data using human activity to tag the content. The video annotation tool used in the present work was the Software ATLAS, a graphical tool for annotating multimodal data flows. The hypothesis of the present work is that it is possible to annotate the task events of a data collection using video and signal track synchronized. The aim of this study is to elucidate the steps to complete the task event annotation process using software ATLAS. The validation of the proposed method was carried out using inertial sensors and video recordings. The biomedical signals included were the signals from the accelerometer and gyroscope during gait. Data collection was filmed with a cellphone camera so that the physical mobility tasks could be watched by researchers to discriminate the task events. The task events were manually annotated, and the inertial signal for each task event was labeled as initial and final time. Data video annotation as proposed here could be used as a tool to properly interpret the results and understand possible inconsistencies that may arise during the data analysis process and to understand the biological system responsible for generating the signals. Results show that the authors were able to annotate the task events of a data collection using the software ATLAS. The seven steps carried out to complete the task event annotation process are elucidated in the present study.