metadata

dataset_info:
  - config_name: corpus
    features:
      - name: corpus-id
        dtype: int64
      - name: image
        dtype: image
      - name: pdf_url
        dtype: string
      - name: company
        dtype: string
      - name: date
        dtype: string
    splits:
      - name: test
        num_bytes: 842829685.81
        num_examples: 1538
    download_size: 761076653
    dataset_size: 842829685.81
  - config_name: qrels
    features:
      - name: query-id
        dtype: int64
      - name: corpus-id
        dtype: int64
      - name: score
        dtype: int64
    splits:
      - name: test
        num_bytes: 3072
        num_examples: 128
    download_size: 2521
    dataset_size: 3072
  - config_name: queries
    features:
      - name: query-id
        dtype: int64
      - name: query
        dtype: string
      - name: source_type
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: test
        num_bytes: 35047
        num_examples: 52
    download_size: 23714
    dataset_size: 35047
configs:
  - config_name: corpus
    data_files:
      - split: test
        path: corpus/test-*
  - config_name: qrels
    data_files:
      - split: test
        path: qrels/test-*
  - config_name: queries
    data_files:
      - split: test
        path: queries/test-*

Vidore Benchmark 2 - ESG Human Labeled

This dataset is part of the "Vidore Benchmark 2" collection, designed for evaluating visual retrieval applications. It focuses on the theme of ESG reports from the fast food industry.

Dataset Summary

Each query is in english.

This dataset provides a focused benchmark for visual retrieval tasks related to ESG reports for the fast food industry. It includes a curated set of documents, queries, relevance judgments (qrels), and page images. This dataset was fully labelled by hand, has no overlap of queries with its synthetic counterpart (available here)

Number of Documents: 27
Number of Queries: 52
Number of Pages: 1538
Number of Relevance Judgments (qrels): 128
Average Number of Pages per Query: 2.5

Dataset Structure (Hugging Face Datasets)

The dataset is structured into the following columns:

corpus: Contains page-level information:
- "image": The image of the page (a PIL Image object).
- "corpus-id": A unique identifier for this specific page within the corpus.
queries: Contains query information:
- "query-id": A unique identifier for the query.
- "query": The text of the query.
qrels: Contains relevance judgments:
- "corpus-id": The ID of the relevant page.
- "query-id": The ID of the query.
- "answer": Answer relevant to the query AND the page.
- "score": The relevance score.

Usage

This dataset is designed for evaluating the performance of visual retrieval systems, particularly those focused on document image understanding.

Example Evaluation with ColPali (CLI):

Here's a code snippet demonstrating how to evaluate the ColPali model on this dataset using the vidore-benchmark command-line tool.

Install the vidore-benchmark package:
```
pip install vidore-benchmark datasets
```

Run the evaluation:

vidore-benchmark evaluate-retriever \
    --model-class colpali \
    --model-name vidore/colpali-v1.3 \
    --dataset-name vidore/restaurant_esg_reports_beir \
    --dataset-format beir \
    --split test

For more details on using vidore-benchmark, refer to the official documentation: https://github.com/illuin-tech/vidore-benchmark

Citation

If you use this dataset in your research or work, please cite:

@misc{faysse2024colpaliefficientdocumentretrieval,
  title={ColPali: Efficient Document Retrieval with Vision Language Models}, 
  author={Manuel Faysse and Hugues Sibille and Tony Wu and Bilel Omrani and Gautier Viaud and Céline Hudelot and Pierre Colombo},
  year={2024},
  eprint={2407.01449},
  archivePrefix={arXiv},
  primaryClass={cs.IR},
  url={https://arxiv.org/abs/2407.01449}, 
}

Acknowledgments

This work is partially supported by ILLUIN Technology, and by a grant from ANRT France.