From October 29 to November 1, 2024, at the <i>2024 IEEE 13th Global Conference on Consumer Electronics (GCCE 2024)</i> held in Kitakyushu, Associate Professor Uchiyama presented the research findings of alumnus Mr. Nakamura and received the Presentation Award.
- Ren Nakamura, Yutaro Hirao, Monica Perusquía-Hernández, Hideaki Uchiyama and Kiyoshi Kiyokawa,
“Generalizing Listening Human Behavior: 3D Human Pose Estimation Using Music”
(Abstract): Existing methods have utilized active acoustic sensing for 3D human pose estimation. However, these approaches are limited in their applicability due to the constraint of using Time Stretched Pulse signals. This paper proposes a novel machine learning-based technique to achieve pose estimation using more common acoustic signals, such as music. Our method employs the differential mel spectrograms of received acoustic signals with and without human presence. Through experimental validation, we demonstrate the potential for human pose estimation even with music not used in the training process, thus indicating the broader applicability of our approach.
(URL): 10.1109/GCCE62371.2024.10760628
