Chest X-Ray Images for Pneumonia Classification

A collection of labeled chest X-ray images for binary classification tasks. The dataset contains 13,977 images split into training, validation, and test sets, with labels for 'PNEUMONIA' and 'NORMAL'. It was uploaded to Hugging Face by user 'trpakov' in March 2023.

Use Cases

• Train a binary image classifier to detect pneumonia based on labeled X-ray images.
• Benchmark model performance on a medical imaging task using the provided train/validation/test splits.
• Fine-tune pre-trained vision models for a specific clinical application.
• Explore data augmentation techniques for a relatively small medical image dataset.

Strengths

• Dataset is pre-split into train (12,230 images), validation (1,165 images), and test (582 images) sets.
• Clear binary labels ('PNEUMONIA' and 'NORMAL') for a focused classification task.

Limitations

• Column-level documentation and sample data are unavailable, making field semantics and data format uncertain.
• The original data source, collection methodology, and patient demographics are not detailed in the provided metadata.
• License information is not explicitly stated; users must follow the link to the dataset page for details.

Provenance

Source

Originally sourced from a Roboflow project (mohamed-traore-2ekkp/chest-x-rays-qjmia).

Collection Method

Method of original image collection and annotation is not described.

Time Range

null

Freshness

Last updated 2023-03 13 07:23:48; freshness should be verified.

Geography

null