ODGS-SLAM dataset

Dataset Overview

This dataset is designed to evaluate and compare visual SLAM systems that use different image modalities. In particular, it enables comparisons between omnidirectional SLAM systems that use multi-camera or panorama inputs and conventional SLAM systems that rely only on perspective cameras.

To support this, the dataset provides virtual and real-world motion sequences for omnidirectional multi-camera, fisheye, and perspective camera setups, together with ground truth trajectories. The motion patterns include isolated directional motion, retracing, forward-backward motion, and more complex exploratory trajectories. Together, these sequences support the evaluation and comparison of visual SLAM systems across different input modalities. The sequences are organized into three environment-based subsets:

• Indoor Room Synthetic:
  • Rendered sequences of a virtual indoor room scene.
  • Data: indoor_room_synthetic.zip
• Indoor Real World:
  • Sequences captured with Insta360Pro and Insta360X4 camera systems mounted on a mobile robot with motion-captured ground truth poses.
  • Data: indoor_real_world.zip
• Outdoor Synthetic:
  • Rendered sequences in a large-scale urban outdoor environment.
  • Data: outdoor_synthetic.zip

All details on available data (e.g., camera modalities, depth maps, and ground-truth trajectories), scene information, and sequence organization are provided in the README.md files, both inside the dataset zip files and outside the zip archives (indoor_room_synthetic_README.md, indoor_real_world_README.md, outdoor_synthetic_README.md).

This dataset was used for the thorough evaluation of "ODGS-SLAM: Omnidirectional Gaussian Splatting SLAM" which is accepted at CVPR 2026 and will be published there. Additional details about the dataset are provided in the paper and supplementary material (see project page: https://odgs-slam.github.io/).

Citation and Attribution

"ODGS-SLAM: Omnidirectional Gaussian Splatting SLAM" has been accepted at CVPR 2026 and will be published in the conference proceedings.
If you use this dataset, please cite the ODGS-SLAM paper once the official reference is available.
Please visit the project page for citation details and the latest paper and project information:

https://odgs-slam.github.io/

@InProceedings{Spiss_2026_CVPR,
    author    = {Spiss, Stefan and Hieronimy, Joey and Ritter, Marcel and Harders, Matthias},
    title     = {{ODGS-SLAM}: {O}mnidirectional {G}aussian {S}platting {SLAM}},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2026},
    pages     = {26114-26123}
}