This repository contains a curated set of high-resolution wildlife photographs from India, each annotated with species, location and descriptive captions. All images were captured with the same camera under consistent settings, making this dataset suitable for computer‐vision tasks such as classification, captioning and species recognition.
Dataset Overview
This India Animals Dataset includes 43 images of nine animal categories from Maharashtra and Kerala:
- boar
- cat
- cormorant
- cow
- dog
- palm squirrel (Funambulus palmarum)
- macaque
- gray langur (Semnopithecus)
- tiger
Every image is accompanied by:
- a human-written caption
- the geographic location where it was taken
- the common name of the species
Repository Structure
.
├── images/
│ ├── boar-001.jpeg
│ ├── cat-001.jpeg
│ ├── ...
│ └── tiger-001.jpeg
├── metadata.csv
└── README.md
metadata.csv
| Column | Description |
|---|---|
| filename | Image file name (e.g. semnopithecus-012.jpeg) |
| animal | Common name of the species |
| location | State in India where the photo was taken (Maharashtra or Kerala) |
| caption | Brief description of the scene |
Image Specifications
All photographs were taken with a Sony ILCE-6600 camera using the following settings:
- Resolution: 6000 × 4000 pixels
- Color space: RGB (Display P3 profile)
Usage
Load metadata and image paths in Python:
import pandas as pd
from PIL import Image
# Load annotations
df = pd.read_csv("metadata.csv")
# Example: open first image and print its caption
row = df.iloc[0]
img = Image.open(f"images/{row.filename}")
print(row.caption)
img.show()
For training on Hugging Face Hub:
from datasets import load_dataset
dataset = load_dataset("NoeFlandre/IndiaAnimals", split="train")
Citation
If you use this dataset, please cite as:
@misc{india-animals-2025,
title = {India Animals Dataset},
author = {Noé Flandre},
year = {2025},
howpublished = {\url{https://huggingface.co/datasets/NoeFlandre/IndiaAnimals}}
}
License
This work is released under a Creative Commons Attribution 4.0 International License (CC BY 4.0). Feel free to use and adapt these images and annotations, provided you give appropriate credit.