erenfazlioglu/turkey-districts-streetviews

Istanbul Districts Image Dataset

A comprehensive dataset of 104,024 street-level images from 39 districts of Istanbul, Turkey. This dataset is designed for machine learning and computer vision applications, particularly for location classification tasks.

Dataset Overview

• Total Images: 104,024
• Districts: 39 districts of Istanbul
• Split: 80% training (83,000 images), 20% testing (21,000 images)
• Top 5 Districts by Image Count:
  • Fatih: 13,103 images
  • Beşiktaş: 11,155 images
  • Eyüpsultan: 10,810 images
  • Üsküdar: 8,993 images
  • Kâğıthane: 8,978 images

Features

Each image in the dataset includes:

• image: The image file
• label: District category (0-38)
• district: District name as string

Usage

Load the dataset using the Hugging Face Datasets library:

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("your-username/repository-name")

# Access train and test sets
train_dataset = dataset["train"]
test_dataset = dataset["test"]

# Access an example
image = train_dataset[0]["image"]
district = train_dataset[0]["district"]

Sample Model Training

Here's a quick example using Vision Transformer for district classification:

from transformers import AutoImageProcessor, AutoModelForImageClassification
from datasets import load_dataset

# Load dataset
dataset = load_dataset("your-username/repository-name")

# Image processor and model
processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224")
model = AutoModelForImageClassification.from_pretrained(
    "google/vit-base-patch16-224", 
    num_labels=39,
    id2label={i: district for i, district in enumerate(dataset["train"].features["district"].names)}
)

# Prepare dataset
def preprocess_data(examples):
    return processor(images=examples["image"], return_tensors="pt")

train_dataset = dataset["train"].map(preprocess_data, batched=True)
# Continue with model training...

Source

This dataset contains street-level images collected from various districts of Istanbul, Turkey. The images were gathered and processed for use in the GeoGuessr project.

License

This dataset is available for educational and research purposes. For commercial use, please contact the author.

Citation

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

@dataset{istanbul_districts_2023,
  author    = {EREN FAZLIOĞLU},
  title     = {Istanbul Districts Image Dataset},
  year      = {2025},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/erenfazlioglu}
}