license: cc-by-nc-4.0
tags:
- person-detection
- action-recognition
- physical-ai
- spatial-reasoning
- spatial-intelligence
pretty_name: SynDORBench
extra_gated_heading: Acknowledge the license to access SynDORBench
extra_gated_description: >-
SynDORBench is released for non-commercial research under CC BY-NC 4.0, for
evaluation only.
extra_gated_prompt: >
By requesting access you agree to use SynDORBench under the CC BY-NC 4.0
license — for non-commercial purposes only, with attribution — to use it
solely for evaluation and benchmarking and NOT to train, fine-tune, or
otherwise fit models on it, and to cite the SynDORBench paper in any resulting
work.
extra_gated_fields:
Full name: text
Affiliation: text
Country: country
Intended research use: text
I agree to the CC BY-NC terms (non-commercial use, with attribution): checkbox
I agree to use this dataset for evaluation only, not to train or fine-tune models on it: checkbox
extra_gated_button_content: Acknowledge and request access
SynDORBench: Evaluating LVLM Perceptual Robustness Under Physically Constrained Visibility Conditions
Figure 1. Conceptual overview of SynDORBench: The benchmark systematically varies physical visibility factors (lighting, scene type, camera azimuth/elevation, action pose) and DORI-calibrated viewing distance, evaluating models on 3 tasks: 1) binary human-presence classification and 2) action recognition (MCQ and open-ended)
Description
SynDORBench is a synthetic benchmark for robust perception under controlled distance and lighting conditions. Scenes span the DORI regions (Detection / Observation / Recognition) distance scale crossed with lighting bands (day / night / night-vision), for fine-grained evaluation of how perception degrades at range and in low light.
Figure 2. Sample images from SynDORBench.
SyndDORBench consists of 87k QA pairs:
- action recognition (open-ended): 27k
- action mcq : 27k
- human-presence classification : 27k + 5k empty scenes
Contents
SynDORBench/
├── images/ # rendered frames + annotations
└── index/
├── 20k_index.jsonl # unified record index across all tasks
├── human_detection/ # human-presence classification task
│ ├── {day,night,nv}.jsonl # GT person-present index
│ ├── empty_{day,night,nv}.jsonl # empty-scene (person-absent) negatives
├── action_recognition/ # open-ended action recognition (similarity-scored)
│ └── {day,night,nv}.jsonl # action-label index
└── action_mcq/ # 8-way multiple-choice action recognition
└── {day,night,nv}.jsonl # MCQ index
Modalities: day = RGB daytime · night = RGB low-light · nv = "night-vision" achieved through image enhancement.
Citation
If you use SynDORBench, please cite the SynDORBench paper.
@article{yee2026syndorbench,
title={SynDORBench: Evaluating LVLM Perceptual Robustness Under Physically Constrained Visibility Conditions},
author={Yee, Jeremy Stephen Gabriel and Wang, Zhengkui and Zhang, Zhiyuan and Anand, Avinash and Liu, Timothy and Chan, Benedict and Ng, Aik Beng and See, Simon},
journal={arXiv preprint arXiv:2609.31823},
year={2026}
}
License
This dataset is released under CC BY-NC 4.0 — non-commercial use, with attribution.