IS3+

IS3+ is an extended version of IS3 with clean audio/image pairs to ensure cross-modality consistency. The dataset has 4 GB of data.

The dataset contains the following data:

• audio_wav: audio files (.wav)
• gt_segmentation:  annotations of image bounding boxes and segmentation masks
• images: images (.jpg)
• IS3_annotation.json: file with image/audio/gt information for every dataset sample.

This work was done as part of the paper Learning from Silence and Noise for Visual Sound Source Localization Models.

Paper citation:

@misc{juanola2025learningsilencenoisevisual,
      title={Learning from Silence and Noise for Visual Sound Source Localization}, 
      author={Xavier Juanola and Giovana Morais and Magdalena Fuentes and Gloria Haro},
      year={2025},
      eprint={2508.21761},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2508.21761}, 
}