SAOM

Author(s)

Mariia Khan, Edith Cowan UniversityFollow

Author Identifiers

Mariia Khan: https://orcid.org/0000-0001-6662-4607

Publication Date

2025

Document Type

Dataset

Publisher

Edith Cowan University

School or Research Centre

School of Science

Description

The SAOM dataset is created for the evaluation of the whole-object semantic segmentation in embodied AI indoor environments. The SAOM dataset is tailored for segmentation in dynamic embodied environments, focusing on interactable objects. It includes 54 object classes, all of which are either `pickupable’, `openable’, or `receptacles`. Unlike static-object datasets, the objects in SAOM can undergo transformations, such as being opened, closed, or moved.

DOI

10.25958/b7p2-z351

Methodology

All data is collected via embodied AI simulator for indoor environments - Ai2Thor.

Start of data collection time period

2022

End of data collection time period

2024

Related Publications

Khan, M., Qiu, Y., Cong, Y., Rosenhahn, B., Abu-Khalaf, J., & Suter, D. (2024, October). Segment any object model (SAOM): Real-to-simulation fine-tuning strategy for multi-class multi-instance segmentation. In IEEE International Conference on Image Processing (ICIP) (pp. 582-588). IEEE. https://doi.org/10.48550/arXiv.2403.10780

File Format(s)

png and json

File Size

2 GB

Creative Commons License

This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Contact

mariia.khan@ecu.edu.au

Citation

Khan, M. (2025). SAOM. Edith Cowan University. https://doi.org/10.25958/b7p2-z351