Abstract
Audio-Visual Semantic Segmentation (AVSS) aims to identify, segment, and classify sound-emitting objects in video frames. Previous Transformer-based AVSS approaches largely inherit design principles from image segmentation models. Recent studies show that these image segmentation models contain redundant components that contribute little to the segmentation performance. Following this insight, we propose Encoder-only Audio-Visual Segmentation (EASE). EASE runs at up to 365 FPS, 3x faster than prior State-of-the-Art (SotA) AVS models at comparable accuracy, and trains in under 11 GPU-hours. Furthermore, we achieve SotA AVSS performance across different backbones and input resolutions. Our results demonstrate that AVSS can be both simpler and faster, providing a scalable foundation for future research and real-time applications. Code, model weights, and samples are available at https://ease-avs.notion.site
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Viertola, I., Iashin, V., Tötterström, S., & Rahtu, E. (2026). Less is More: Encoder-only Audio-Visual Segmentation. https://omanscience.com/en/articles/less-is-more-encoder-only-audio-visual-segmentation
MLA 9
Viertola, Ilpo, et al. "Less is More: Encoder-only Audio-Visual Segmentation." https://omanscience.com/en/articles/less-is-more-encoder-only-audio-visual-segmentation.
Chicago (author–date)
Viertola, Ilpo, Vladimir Iashin, Sophie Tötterström, and Esa Rahtu. 2026. "Less is More: Encoder-only Audio-Visual Segmentation." https://omanscience.com/en/articles/less-is-more-encoder-only-audio-visual-segmentation.
Harvard
Viertola, I., Iashin, V., Tötterström, S. and Rahtu, E. (2026) 'Less is More: Encoder-only Audio-Visual Segmentation', Available at: https://omanscience.com/en/articles/less-is-more-encoder-only-audio-visual-segmentation.
Vancouver
Viertola I, Iashin V, Tötterström S, Rahtu E. Less is More: Encoder-only Audio-Visual Segmentation. https://omanscience.com/en/articles/less-is-more-encoder-only-audio-visual-segmentation
IEEE
I. Viertola, V. Iashin, S. Tötterström, and E. Rahtu, "Less is More: Encoder-only Audio-Visual Segmentation," https://omanscience.com/en/articles/less-is-more-encoder-only-audio-visual-segmentation.