Visually Guided Sound Source Separation and Localization using Self-Supervised Motion Representations.
Lingyu ZhuEsa RahtuPublished in: CoRR (2021)
Keyphrases
- visually guided
- source separation
- blind source separation
- temporal structure
- denoising
- independent component analysis
- motion estimation
- image sequences
- human motion
- obstacle avoidance
- camera motion
- motion patterns
- audio features
- single channel
- moving objects
- image classification
- humanoid robot
- language model
- spatio temporal
- computer vision