Fig. 1From: First-person reading activity recognition by deep learning with synthetically generated imagesOverview of our approachBack to article page