0xRafie/asean-geolocation

ASEAN Geolocation

ASEAN Geolocation is an image dataset for country-level geolocation across the 11 ASEAN countries. It is designed for image classification, visual geolocation, and explainable geolocation research.

The dataset contains 4850 images organized by country label. Images are provided at their original aspect ratios; model-specific resizing and normalization should be handled during preprocessing.

Dataset summary

• Task: country-level image classification / image geolocation
• Countries: 11 ASEAN countries
• Images: 4850
• Labels: ISO-style country code, country name, and integer class label
• Image format: mixed original image formats and aspect ratios

Dataset composition

The public dataset is organized by country. Public filenames are source-free to avoid filename-based source shortcuts.

Country code	Country	Images
TH	Thailand	500
SG	Singapore	500
MY	Malaysia	500
ID	Indonesia	500
PH	Philippines	500
KH	Cambodia	350
MM	Myanmar	350
LA	Laos	500
VN	Vietnam	500
BN	Brunei	300
TL	Timor-Leste	350
Total		4850

Data fields

The public metadata file is metadata.csv with the following columns:

Column	Type	Description
file_name	string	Relative path to the image file.
country_code	string	Country code label.
country_name	string	Human-readable country name.
label	integer	Integer class label.

The current label order is:

Label	Country code	Country
0	TH	Thailand
1	SG	Singapore
2	MY	Malaysia
3	ID	Indonesia
4	PH	Philippines
5	KH	Cambodia
6	MM	Myanmar
7	LA	Laos
8	VN	Vietnam
9	BN	Brunei
10	TL	Timor-Leste

Directory structure

README.md
metadata.csv
images/
  BN/
    BN_000001.jpg
    ...
  ID/
    ID_000001.jpg
    ...
  ...

Intended uses

• Country-level image geolocation.
• Image classification benchmarks for ASEAN countries.
• Explainable geolocation experiments.
• Visual geography and regional representation studies.

Out-of-scope uses

• Identifying exact image locations.
• Surveillance, person identification, or face recognition.
• Demographic or sensitive-attribute inference.
• Legal, safety-critical, or high-stakes decision making.

Limitations and biases

This dataset is a curated benchmark, not an exhaustive or statistically representative sample of each country. Images may overrepresent urban, roadside, tourist, or web-visible scenes. Visual domain differences may remain even though public filenames and public metadata do not expose per-image source information.

The dataset is intended for country-level geolocation research. It should not be interpreted as a complete representation of people, places, cultures, or environments in any country.

Loading example

from pathlib import Path
import pandas as pd
from PIL import Image

root = Path("hf_publish/asean-geolocation")
metadata = pd.read_csv(root / "metadata.csv")

row = metadata.iloc[0]
image = Image.open(root / row["file_name"])
label = row["country_code"]

Citation

If you use this dataset, please cite it as:

@dataset{ekaputra_asean_geolocation_2026,
  title     = {ASEAN Geolocation},
  author    = {Ekaputra, Danish Rafie},
  year      = {2026},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/0xRafie/asean-geolocation},
  note      = {Country-level image geolocation dataset covering 11 ASEAN countries}
}

License and usage

This dataset is provided for research and educational use as a curated benchmark for country-level image geolocation. The metadata, labels, and dataset organization are provided by Danish Rafie Ekaputra.

The image files were assembled from mixed public and street-level imagery sources and may remain subject to their original copyrights, licenses, and platform terms. Users are responsible for ensuring that their use complies with applicable laws, source terms, and institutional policies.

Do not use this dataset for surveillance, person identification, demographic inference, or safety-critical decision making.