zjunlp/OceanBenchMark

---
dataset_info:
features:
- name: input
dtype: string
- name: output
dtype: string
- name: image_path
dtype: string
splits:
- name: test
num_bytes: 0
num_examples: 0
configs:
- config_name: science_qa
description: "Marine Science QA (Text only)."
- config_name: science_vqa
description: "Marine Science VQA."
- config_name: sonar_vqa
description: "Sonar Image QA."
- config_name: organism_vqa
description: "Marine Organism VQA."
license: mit
language:
- zh
- en
task_categories:
- question-answering
- image-text-to-text
pretty_name: OceanBenchmark
tags:
- benchmark
- evaluation
- ocean
- marine-science
---

# OceanBenchmark

## 1. Dataset Description

OceanBenchmark is a benchmark dataset designed to evaluate the comprehensive capabilities of marine-focused large models. It encompasses a diverse range of tasks, spanning from unimodal marine science knowledge question answering to complex multimodal visual question answering.

## 2. Sub-datasets

| Subset Directory | Task Type | Sample Size | Description |
|:---|---|---|---|
| **Ocean Science QA** | QA | 102 | Text-only multiple-choice and open-ended questions in marine science. |
| **Ocean Science VQA** | VQA | 99 | Visual question answering based on scientific diagrams and imagery. |
| **Sonar VQA Marine** | VQA | 796 | Target detection and question answering evaluation on sonar imagery. |
| **Marine Organisms VQA** | VQA | 472 | Classification and identification tests for marine organisms. |

## 3. Usage Example

```python
from datasets import load_dataset
```

# Load the sonar evaluation subset
ds_test = load_dataset("zjunlp/OceanBenchmark", "sonar_vqa", split="test")
print(ds_test[0]['input'])

## 4. License
License: MIT