Key Points
To account for differences in subjects' poses and body shapes, we map gaze distributions onto a neutral human body model with a fixed pose and shape. Three-dimensional heatmaps on the model's surface make it easier to compare where observers' gaze concentrates.
Abstract
When we look at a photograph of a person, our eyes move between places such as the face and clothing. A heatmap can use colors to show where viewers looked most. However, people in different photographs have different poses and body shapes. This makes it difficult to compare heatmaps placed directly on the original photographs.
We map the measured gaze locations onto a shared three-dimensional human body model with a fixed pose and shape. Displaying heatmaps on this model puts the gaze measurements on a common basis. Researchers can then compare which body regions attracted attention across photographs without having to account for a different pose and shape each time.
We compared this approach with the conventional method of placing a flat heatmap over each photograph. Experiments showed that the shared body model allowed researchers studying eye movements to compare gaze distributions more directly.
Publications
- Michiko Inoue, Fuyuko Iwasaki, Masashi Nishiyama, Yoshio Iwai,
Heatmap overlay using neutral body model for visualizing the measured gaze distributions of observers, Proceedings of 7th Asian Conference on Pattern Recognition (ACPR), Part II, pp. 102 - 114, November 2023. [Slides] - 岩崎 芙由子, 井上 路子, 米田 駿介, 西山 正志,
被写体動作の躍動感を評価する観察者から計測された視線の3Dヒートマップ表現,
信学技報, vol. 123, no. 358, PRMU2023-39, pp. 1 - 6, January 2024. [Slides] [Short movie]
