The Sound of Motions
Author(s)
Zhao, Hang; Gan, Chuang; Ma, Wei-Chiu; Torralba, Antonio
DownloadSubmitted version (5.928Mb)
Open Access Policy
Open Access Policy
Creative Commons Attribution-Noncommercial-Share Alike
Terms of use
Metadata
Show full item recordAbstract
© 2019 IEEE. Sounds originate from object motions and vibrations of surrounding air. Inspired by the fact that humans is capable of interpreting sound sources from how objects move visually, we propose a novel system that explicitly captures such motion cues for the task of sound localization and separation. Our system is composed of an end-to-end learnable model called Deep Dense Trajectory (DDT), and a curriculum learning scheme. It exploits the inherent coherence of audio-visual signals from a large quantities of unlabeled videos. Quantitative and qualitative evaluations show that comparing to previous models that rely on visual appearance cues, our motion based system improves performance in separating musical instrument sounds. Furthermore, it separates sound components from duets of the same category of instruments, a challenging problem that has not been addressed before.
Date issued
2019-10Department
MIT-IBM Watson AI Lab; Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory; Massachusetts Institute of Technology. Department of Electrical Engineering and Computer ScienceJournal
Proceedings of the IEEE International Conference on Computer Vision
Publisher
IEEE
Citation
Zhao, Hang, Gan, Chuang, Ma, Wei-Chiu and Torralba, Antonio. 2019. "The Sound of Motions." Proceedings of the IEEE International Conference on Computer Vision, 2019-October.
Version: Original manuscript