Discriminative Sounding Objects Localization via Self-supervised Audiovisual Matching
Abstract
Discriminatively localizing sounding objects in cocktail-party, i.e., mixed sound scenes, is commonplace for humans, but still challenging for machines. In this paper, we propose a two-stage learning framework to perform self-supervised class-aware sounding object localization. First, we propose to learn robust object representations by aggregating the candidate sound localization results in the single source scenes. Then, class-aware object localization maps are generated in the cocktail-party scenarios by referring the pre-learned object knowledge, and the sounding objects are accordingly selected by matching audio and visual object category distributions, where the audiovisual consistency is viewed as the self-supervised signal. Experimental results in both realistic and synthesized cocktail-party videos demonstrate that our model is superior in filtering out silent objects and pointing out the location of sounding objects of different classes. Code is available at https://github.com/DTaoo/Discriminative-Sounding-Objects-Localization.
Cite
@article{arxiv.2010.05466,
title = {Discriminative Sounding Objects Localization via Self-supervised Audiovisual Matching},
author = {Di Hu and Rui Qian and Minyue Jiang and Xiao Tan and Shilei Wen and Errui Ding and Weiyao Lin and Dejing Dou},
journal= {arXiv preprint arXiv:2010.05466},
year = {2020}
}
Comments
To appear in NeurIPS 2020. Previous Title: Learning to Discriminatively Localize Sounding Objects in a Cocktail-party Scenario