In the Wild Settings

AbstractExisting video understanding datasets mostly focus on human interactions, with little attention being paid to the “in the wild” settings, where the videos are recorded outdoors. We propose WILDQA, a …...