Image Censorship Dataset for NSFW Content Detection Benchmarking

An image dataset compiled for benchmarking NSFW content detection models. The dataset is a curated compilation of existing images with labels for safe and unsafe content, designed to have a distribution similar to real user input or generated output. It was created by author vekshinkir and last updated on June 5, 2026.

Use Cases

• Benchmarking NSFW detection models based on the labeled safe/unsafe images.
• Training classifiers for content moderation based on the provided image categories.
• Analyzing edge cases in content moderation based on the 'is_edge_case' flag for examples like 'sexy' or 'spa'.
• Studying dataset distribution for realistic user or generated content simulation.

Strengths

• Includes explicit labels for 'safe' and 'unsafe' content categories.
• Contains a boolean flag ('is_edge_case') to identify borderline examples for nuanced analysis.
• Designed with a distribution intended to mimic real-world user input or generated output.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license information are unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Curated compilation of existing datasets with added labeling.

Time Range

null

Freshness

Last updated 2026-06-05 20:27:12; freshness should be verified.

Geography

null