Detect Artificial Texts and Watermarks in Images and Videos

The Text Detection endpoint is helpful to detect specific texts that have been artificially added (through post-processing) to Images and Videos, for instance company names, phone numbers, personal information or emails.

Discriminate between artificial and natural text

Our model has been trained to discriminate between natural text (i.e. text that naturally occurs in photos such as on buildings, road signs etc) and artificial text that has been added in post-processing.

All languages, all fonts

The model was trained to detect text in all orientations, alphabets (latin, chinese, japanese, korean, arabic, hebrew etc) and for all types of fonts from traditional to cursive.


Why use Sightengine?

Fast and scalable

The API returns moderation results instantly and scales automatically to adapt to your needs.
Easily grow your Moderation Pipeline to tens of millions of images per month.


Built upon state-of-the-art models and proprietary technology. The moderation decisions are consistent and auditable, with feedback loops and continuous improvement built-in.

Easy to integrate

The API was built by developers for developers. You only need a few lines of code to be up and running.
Leverage our simple SDKs and detailed documentation.

Privacy compliant

No human moderator is involved, your images remain private and are not shared with any 3rd party.


Check out our other models

Standard Moderation

Face / People related

Text analysis, QR codes and OCR

AI Generation and Editing

Content Description

Near-duplicate Detection

Fraud Detection


Ready to get a taste of the future?

Empower your Business with powerful Multimodal Models. We power hundreds of platforms and apps worldwide.