IndicPope: Indian Multilingual Translation Dataset For Evaluating Large Vision Language Models

• You can find the performance of Chitrarth on IndicPope here : Paper | Github | HuggingFace
• Evaluation Scripts of BharatBench is available here : Github

1. Introduction

IndicPope is a new dataset designed for evaluating Large Vision-Language Models (LVLMs) on Visual Question Answering (VQA) tasks. It focuses on simple Yes-or-No questions probing objects in images (e.g., Is there a car in the image?).

This dataset is built upon POPE: Polling-based Object Probing Evaluation for Object Hallucination (GitHub), which employs negative sampling techniques to test hallucination in vision-language models under Random, Popular, and Adversarial settings.

---

2. Dataset Details

IndicPope consists of 8.91k samples spanning 12 Indic languages along with English. Each sample includes:

• Text: The question about the image.
• Category: The type of sampling used (Random/Popular/Adversarial).
• Label: The answer (Yes/No).

Supported Languages

• Assamese
• Bengali
• English
• Gujarati
• Hindi
• Kannada
• Malayalam
• Marathi
• Odia
• Sanskrit
• Tamil
• Telugu

---

3. How to Use and Run

You can load the dataset using the datasets library:

from datasets import load_dataset

dataset = load_dataset("krutrim-ai-labs/IndicPope")
print(dataset)

---

4. License

This code repository and the model weights are licensed under the Krutrim Community License.

5. Citation

@article{khan2025chitrarth,
  title={Chitrarth: Bridging Vision and Language for a Billion People},
  author={Shaharukh Khan, Ayush Tarun, Abhinav Ravi, Ali Faraz, Akshat Patidar, Praveen Kumar Pokala, Anagha Bhangare, Raja Kolla, Chandra Khatri, Shubham Agarwal},
  journal={arXiv preprint arXiv:2502.15392},
  year={2025}
}

@misc{liu2023improvedllava,
      title={Improved Baselines with Visual Instruction Tuning}, 
      author={Liu, Haotian and Li, Chunyuan and Li, Yuheng and Lee, Yong Jae},
      publisher={arXiv:2310.03744},
      year={2023},
}

@misc{liu2023llava,
      title={Visual Instruction Tuning}, 
      author={Liu, Haotian and Li, Chunyuan and Wu, Qingyang and Lee, Yong Jae},
      publisher={NeurIPS},
      year={2023},
}

@article{li2023evaluating,
  title={Evaluating object hallucination in large vision-language models},
  author={Li, Yifan and Du, Yifan and Zhou, Kun and Wang, Jinpeng and Zhao, Wayne Xin and Wen, Ji-Rong},
  journal={arXiv preprint arXiv:2305.10355},
  year={2023}
}

@article{gala2023indictrans2,
  title={Indictrans2: Towards high-quality and accessible machine translation models for all 22 scheduled indian languages},
  author={Gala, Jay and Chitale, Pranjal A and AK, Raghavan and Gumma, Varun and Doddapaneni, Sumanth and Kumar, Aswanth and Nawale, Janki and Sujatha, Anupama and Puduppully, Ratish and Raghavan, Vivek and others},
  journal={arXiv preprint arXiv:2305.16307},
  year={2023}
}

6. Contact

Contributions are welcome! If you have any improvements or suggestions, feel free to submit a pull request on GitHub.

7. Acknowledgement

IndicPope is built with reference to the code of the following projects: POPE, and LLaVA-1.5. Thanks for their awesome work!