Challenges in using wearable cameras and phone logs to generate ground truth of transportation activities.

Researchers have started recruiting users to record activity data with ground truth annotations in real-word settings. These annotations are considered "ground truth" because users who collected activity data probably know what happened during the activity. However, while ground truth annotations are generally believed to be truthful representations of recorded activities, other qualities of activity recordings such as temporal precision are also of researchers' interest. To assess and compare the qualities of annotated recordings collected among different approaches, it is necessary to have another way to capture a ground truth of users' activities. In this paper, we demonstrate how we used wearable cameras and phone logs to assess the quality of users' recorded transportation activity data. We focus on how we collected photos and logs and generated a ground truth of transportation activity data, and we present challenges we encountered in the generation process.