Nsfwjs

Nsfwjs is an AI Content Filtering Tool designed to check images for indecent content without sending them to a server. All analysis runs locally on your device, helping you keep image data private while you evaluate whether an image should be flagged.

The tool is powered by a model that loads in the browser and reports detection accuracy levels such as 90% (about 2.6MB), 93% (about 4.2MB), and a larger “Huge” option at 93% accuracy. You’ll see a “Ready to Classify” state once the model is loaded, indicating the checker is available to classify images.

For input, Nsfwjs supports drag-and-drop image checking and also includes a Camera option for capturing images directly. It also features Blur Protection to help handle images where privacy or legibility protections are needed before classification.

Nsfwjs is best for developers and product teams who want lightweight, privacy-focused image moderation workflows, as well as for anyone experimenting with client-side NSFW detection. GitHub resources are available for the code, the website, the NSFW model, and a mobile demo, making it easier to explore integration paths.

Does Nsfwjs send images to a server for checking?

No—Nsfwjs performs indecent content checking entirely on the client side, so images are processed locally on your device.

How accurate is Nsfwjs at detecting indecent content?

It offers reported accuracy levels such as 90% (around 2.6MB) and 93% (around 4.2MB, and a “Huge” model also at 93%).

What ways can I provide an image to Nsfwjs?

You can drag and drop an image to check, or use the Camera option to capture an image for classification.

What is Blur Protection in Nsfwjs?

Blur Protection is a feature designed to help with how images are handled for classification, particularly when dealing with blurred content before detection.

Is Nsfwjs usable in mobile apps?

Yes—there’s a mobile demo available on GitHub to show how the approach can work in mobile environments.

Where can I find the code and model resources?

You can access GitHub links for the NSFW.JS project, the website, the NSFW model, and a mobile demo, plus a blog post for additional context.