Sign in to view source links and access this dataset
Description
UFO-MNIST is a dataset of 10,000 28x28 grayscale images depicting UFO-like spotting patterns and common aerial lookalikes. It was created by author 'tentime' and last updated on Hugging Face in May 2026. The dataset is structured as a compact benchmark with 8,000 training and 2,000 test examples across 10 labels.
Use Cases
Benchmarking image classifier performance based on the described 10-class aerial image task.
Demonstrating machine learning workflows using the compact, MNIST-style format mentioned in the description.
Training models to distinguish between UFO-like patterns and aerial lookalikes as described.
Educational tutorials on computer vision using a small, well-structured image dataset.
Strengths
Provides a defined benchmark structure with 8,000 training and 2,000 test examples.
Images are in a consistent, lightweight 28x28 grayscale format suitable for rapid prototyping.
Includes 10 distinct labels for a multi-class classification problem.
Limitations
Description metadata is limited; actual data quality and image content require manual inspection after download.
Column-level documentation is absent; field semantics must be inferred after download.
The dataset's small scale of 10,000 images may limit its use for training complex, modern models.
Provenance
Source
Hugging Face dataset by author 'tentime'.
Collection Method
Collection method is not specified in the provided description.
Time Range
Temporal coverage is not specified.
Freshness
Last updated 2026-05-25 21:45:01; freshness should be verified.
Geography
Spatial coverage is not specified.
License is unknown; users must verify the license terms on the dataset page before use.