REVEAL: A Large-Scale Image Dataset for Steganalysis

100,006 original images taken with more than 50 different cameras, ranging in size from 256x256 to 7680x4320. The REVEAL dataset was created by M. Kombrink and includes over 300,000 images after preprocessing and applying more than 50 steganography algorithms. This annotated dataset is intended for developing and validating detection methods for law enforcement and academic research.

Use Cases

• Training supervised machine-learning models for steganography detection based on the annotated original, preprocessed, and stego image sets.
• Forensic evaluation and validation of detection algorithms based on the dataset's wide distribution of image attributes.
• Benchmarking new steganography detection methods against a large-scale representative dataset with many preprocessing chains.

Strengths

• Contains 100,006 original images captured with more than 50 different cameras.
• Includes over 300,000 total images after preprocessing and steganography application.
• Features a wide distribution of image attributes, including over 200 different sizes.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for individual sets is unknown, which may limit suitability assessment.
• Last updated 2026-07-04 10:13:20; freshness should be verified.

Provenance

Source

DANS Data Station Physical and Technical Sciences Collection

Collection Method

Images were captured, subjected to preprocessing chains, and then information was hidden using steganography algorithms.

Freshness

Last updated 2026-07-04 10:13:20.