IAS-160k-C

Dataset description

IAS-160k-C is a large-scale benchmark dataset for fine-grained visual classification of invasive alien species, supporting research at the intersection of computer vision and ecological conservation. It contains 162,698 species-labelled images covering 51 representative invasive alien species, including both plants and animals, and is intended for training, validating, and testing deep learning models for species recognition.

The dataset is divided into training, validation, and test subsets at an approximate ratio of 70:15:15:

Split	Images	Percentage
Training	113,860	70.0%
Validation	24,381	15.0%
Test	24,457	15.0%
Total	162,698	100.0%

ScienceDB mirror

This repository is a mirror of IAS-160k-C, available on ScienceDB at DOI: 10.57760/sciencedb.32356.

Usage

The original ZIP archive is available for download. Extract it to access the images and original split indices. index/train.txt, index/val.txt, and index/test.txt contain image paths relative to images/, followed by integer class IDs; index/labels.txt provides the class mapping.

The Parquet files support online image browsing and loading with the Hugging Face datasets library:

pip install datasets pillow

from datasets import load_dataset

ds = load_dataset("hf-gx/IAS-160K-C")
sample = ds["train"][0]
image = sample["image"]
class_id = sample["label"]
species = sample["species"]

For streaming access without downloading the complete dataset:

train = load_dataset("hf-gx/IAS-160K-C", split="train", streaming=True)
sample = next(iter(train))

The available splits are train, validation, and test; validation corresponds to the original val split. The Parquet version preserves the original image bytes, class IDs, and split membership, without resizing or re-encoding images.

Images retain their original licenses. Consult the per-image records in source.json for source, attribution, and license information. Author-created annotations use CC BY-NC 4.0; see LICENSE.md.