Multiple Sound Sources Localization from Coarse to Fine

Rui Qian, Di Hu, Heinrich Dinkel, Mengyue Wu, Ning Xu, Weiyao Lin

ECCV 2020

doi:10.1007/978-3-030-58565-5_18 /eccv/2020/qian2020eccv-multiple/

Abstract

How to visually localize multiple sound sources in unconstrained videos is a formidable problem, especially when lack of the pairwise sound-object annotations. To solve this problem, we develop a two-stage audiovisual learning framework that disentangles audio and visual representations of different categories from complex scenes, then performs cross-modal feature alignment in a coarse-to-fine manner. Our model achieves state-of-the-art results on public dataset of localization, as well as considerable performance on multi-source sound localization in complex scenes. We then employ the localization results for sound separation and obtain comparable performance to existing methods. These outcomes demonstrate our model's ability in effectively aligning sounds with specific visual sources.

PDF ECCV Semantic Scholar

Cite

Text

Qian et al. "Multiple Sound Sources Localization from Coarse to Fine." Proceedings of the European Conference on Computer Vision (ECCV), 2020. doi:10.1007/978-3-030-58565-5_18

Markdown

[Qian et al. "Multiple Sound Sources Localization from Coarse to Fine." Proceedings of the European Conference on Computer Vision (ECCV), 2020.](https://mlanthology.org/eccv/2020/qian2020eccv-multiple/) doi:10.1007/978-3-030-58565-5_18

BibTeX

@inproceedings{qian2020eccv-multiple,
  title     = {{Multiple Sound Sources Localization from Coarse to Fine}},
  author    = {Qian, Rui and Hu, Di and Dinkel, Heinrich and Wu, Mengyue and Xu, Ning and Lin, Weiyao},
  booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
  year      = {2020},
  doi       = {10.1007/978-3-030-58565-5_18},
  url       = {https://mlanthology.org/eccv/2020/qian2020eccv-multiple/}
}