Yuta Segawa, Kazuhiko Kawamoto, and Kazushi Okamoto, First-person reading activity recognition by deep learning with synthetically generated images, EURASIP Journal on Image and Video Processing, 2018:33, 2018.