Dataset · truemanv5666
GeoText1652 multimodal geolocalization dataset
truemanv5666/GeoText1652_Dataset
Official image-text dataset for the GeoText1652 research paper containing 100,000 to 1,000,000 records.
What's inside
The dataset contains between 100,000 and 1,000,000 image-text description pairs. This is the official repository for the GeoText1652 research publication (arXiv:2311.12751), though detailed field composition and languages are not specified at the top of the card.
What you can do with it
- Training multimodal models to match terrain images with text descriptions
- Reproducing experiments and benchmarks from the GeoText1652 paper
- Size
- 100K–1M rows
- Downloads this month
- 19.7K
- Likes
- 2
How to load it
from datasets import load_dataset
ds = load_dataset("truemanv5666/GeoText1652_Dataset")