README.md

---
license: cc-by-4.0
task_categories:
- visual-question-answering
- image-text-to-text
tags:
- agriculture
- land-cover
- land-use
- remote-sensing
- vqa
- vision-language
pretty_name: AgriBench
configs:
- config_name: train
  data_files:
  - split: train
    path: train-0000-of-0001.parquet
dataset_info:
  config_name: train
  splits:
  - name: train
    num_examples: 7136
size_categories:
- 1K<n<10K
---

# AgriBench

AgriBench is the first benchmark built to evaluate multimodal LLMs on agriculture tasks. This repo is its MM-LUCAS subset: built on top of the EU's LUCAS (Land Use/Cover Area frame Survey) land-cover survey, it pairs 1,784 landscape photos (1600×1200, taken across 27 EU countries) with microdata the survey already recorded — GPS location, country, capture date, land-cover and land-use taxonomy codes, and quality/aesthetic ratings.

Paper: [AgriBench: A Hierarchical Agriculture Benchmark for Multimodal Large Language Models](https://doi.org/10.1007/978-3-031-91835-3_14) (arXiv:2412.00465, ECCV 2024 Workshops)
Authors: Yutong Zhou, Masahiro Ryo
Original repository: [github.com/Yutong-Zhou-cv/AgriBench](https://github.com/Yutong-Zhou-cv/AgriBench)

Underlying LUCAS survey data:
- [Semantic segmentation dataset of Land Use/Cover Area frame Survey (LUCAS) rural landscape Street View Images](https://doi.org/10.1016/j.dib.2024.110394) — Martinez-Sanchez, Hufkens, Kearsley, Naydenov, Czucz, van de Velde (Data in Brief, 2024)
- [Harmonised LUCAS in-situ land cover and use database for field surveys from 2006 to 2018 in the European Union](https://doi.org/10.1038/s41597-020-00675-z) — d'Andrimont et al. (Scientific Data, 2020)

## About the paper

AgriBench turns the LUCAS microdata into four VQA tasks over the same 1,784 images:

- **Land cover** — 4-option multiple choice over LUCAS's land-cover classes (e.g. "Shrubland with sparse tree cover", "Broadleaved woodland").
- **Land use** — 4-option multiple choice over LUCAS's socio-economic land-use classes (e.g. "Kitchen garden", "Semi-natural and natural areas not in use").
- **Aesthetic score** — a 5-point qualitative rating of the photo itself (Bad / Poor / Fair / Good / Excellent).
- **Quality score** — a 5-point qualitative rating of image/capture quality (Bad / Poor / Fair / Good / Excellent).

License: CC BY 4.0.

## Dataset Size

| Config | Rows | Unique Images | Image Data |
|---|---|---|---|
| train | 7,136 | 1,784 | 0.92GB (1 shard) |

All four tasks share the exact same 1,784 images, so every image contributes 4 rows (one per task) rather than being split into separate configs.

## Task Breakdown

| Task | Mechanism | Rows |
|---|---|---|
| aesthetics_score | open-ended rating (Bad/Poor/Fair/Good/Excellent) | 1,784 |
| quality_score | open-ended rating (Bad/Poor/Fair/Good/Excellent) | 1,784 |
| land_cover | multiple-choice (4 options) | 1,784 |
| land_use | multiple-choice (4 options) | 1,784 |

For the multiple-choice tasks, the assistant turn combines the answer key with its option text, e.g. `"D. Broadleaved woodland"`, rather than the bare letter the source JSON gives. For the rating tasks, the source answer is already a plain-text label (e.g. `"Poor"`), used as-is.

## Layout

```
train-0000-of-0001.parquet          # 7,136 rows
images/
  image-shared-000-of-001.zip       # image shard (ZIP_STORED, uncompressed)
  path_to_shard.parquet             # maps image SHA256 hash -> shard_file
```

Images are content-addressed by SHA256 hash and stored uncompressed in zip shards for random access via `zipfile`, capped at 5GB per shard (this dataset only needed one).

## Schema

| Column | Type | Description |
|---|---|---|
| `images` | `list<struct<bytes, path>>` | `bytes` is `null`; `path` is the SHA256 hash, looked up in `images/path_to_shard.parquet` |
| `id` | `string` | `mmlucas_<task>_<n>` (names the MM-LUCAS subset this repo standardizes) |
| `messages` | chat-style list | user turn: image + question (multiple-choice questions include lettered options inline); assistant turn: the answer, letter-prefixed for multiple-choice questions |
| `raw_metadata` | JSON string | `task`, `image_path` (original source path), `gps_long`, `gps_lat`, `date`, `nuts0` (EU country code), `Classes` (free-text scene description); plus the task-specific field: `Aesthetic Score`, `Quality Score`, `lc1`/`lc1_label`, or `lu1`/`lu1_label` |

Each row is one (image, task) pair, so every image appears across 4 rows.

## Usage

Recommended, via the AgML python library:

```python
from agml import loadImageTextToTextDataset

ds = loadImageTextToTextDataset("AgriBench")
```

## Citations

```bibtex
@article{zhou2024agribench,
  title={AgriBench: A Hierarchical Agriculture Benchmark for Multimodal Large Language Models},
  author={Zhou, Yutong and Ryo, Masahiro},
  journal={arXiv preprint arXiv:2412.00465},
  year={2024}
}

@article{martinez2024semantic,
  title={Semantic segmentation dataset of Land Use/Cover Area frame Survey (LUCAS) rural landscape Street View Images},
  author={Martinez-Sanchez, Laura and Hufkens, Koen and Kearsley, Elizabeth and Naydenov, Dimitar and Cz{\'u}cz, B{\'a}lint and van de Velde, Marijn},
  journal={Data in Brief},
  volume={54},
  pages={110394},
  year={2024},
  publisher={Elsevier}
}

@article{d2020harmonised,
  title={Harmonised LUCAS in-situ land cover and use database for field surveys from 2006 to 2018 in the European Union},
  author={d’Andrimont, Rapha{\"e}l and Yordanov, Momchil and Martinez-Sanchez, Laura and Eiselt, Beatrice and Palmieri, Alessandra and Dominici, Paolo and Gallego, Javier and Reuter, Hannes Isaak and Joebges, Christian and Lemoine, Guido and others},
  journal={Scientific data},
  volume={7},
  number={1},
  pages={352},
  year={2020},
  publisher={Nature Publishing Group UK London}
}
```

---

This dataset is indexed and structured on https://project-agml.github.io/ as part of the AgML python library. This dataset was reformatted from its original format to match HuggingFace's Imagefolder standards but requires an external module (agml) that processes and returns a HF Dataset object faster than HF module functions.