cvssp/OmniCount-191

---
license: openrail++
dataset_info:
features:
- name: image
dtype: image
- name: objects
dtype: string
- name: counts
dtype: string
- name: points
dtype: string
- name: bboxes
dtype: string
splits:
- name: test
num_examples: 30230
---

# OmniCount-191

A comprehensive benchmark for multi-label object counting, introduced in **OmniCount: Multi-label Object Counting with Semantic-Geometric Priors** (AAAI 2025).

## Dataset Description

OmniCount-191 consists of **30,230 images** with multi-label object counts, including **point**, **bounding box**, and **VQA** annotations across 191 object categories.

- **Paper:** [arXiv:2403.05435](https://arxiv.org/abs/2403.05435)
- **Project Page:** [mondalanindya.github.io/OmniCount](https://mondalanindya.github.io/OmniCount/)
- **Code:** [github.com/mondalanindya/OmniCount](https://github.com/mondalanindya/OmniCount)

## Citation
```bibtex
@inproceedings{mondal2025omnicount,
title={OmniCount: Multi-label Object Counting with Semantic-Geometric Priors},
author={Mondal, Anindya and Nag, Sauradip and Zhu, Xiatian and Dutta, Anjan},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
year={2025}
}
```

## License

This dataset is released under the [Open RAIL++-S License](https://huggingface.co/blog/open_rail). Usage that enables monitoring of individuals without proper legal safeguards and ethical constraints is prohibited.