SilverAIDetect text regions and return a mask, ready to feed the Remove Text API.
Text Detection by SilverAI locates text regions in an image and returns a precise mask covering every detected character and word. Rather than altering the image, it produces a pixel-accurate RGBA mask that marks exactly where text appears, leaving you in full control of what happens next. The mask can be reviewed, edited by hand, or fed directly into a downstream removal step.
This model is the natural companion to the Remove Text API. The mask returned here can be passed as the input_mask to Remove Text, optionally after manual touch-ups, giving you a clean, two-stage pipeline: detect first, then remove. Because detection and removal are decoupled, you gain transparency and fine-grained control over which text gets erased, retouched, or preserved.
Multi-language detection: Reliably detects both Latin and non-Latin scripts, including CJK, Cyrillic, Arabic, and more.
Tight, accurate masks: Returns pixel-precise RGBA masks that hug character edges for clean downstream results.
Pairs with Remove Text: The output mask drops straight into the Remove Text API as input_mask for an end-to-end removal pipeline.
Editable output: Masks can be manually refined before removal, so you decide exactly what stays and what goes.
Flexible input: Accepts an uploaded image file or a remote image URL.
Clear empty signal: Returns a null mask and detected: false when no text is present, making conditional logic simple.
Detect text on signs, documents, and labels to build masks that drive automated cleanup. Pair the detected mask with Remove Text to erase captions, watermarks, or overlaid wording while preserving the underlying scene.
With support for both Latin and non-Latin scripts, Text Detection prepares masks for global content workflows. Localize imagery by detecting source-language text before replacing or removing it for new markets.
Standardize product catalogs by detecting promotional text, watermarks, and stray labels across thousands of images. Use the masks to flag images for cleanup or feed them directly into a removal pipeline.
Save up to 70% vs direct pricing
Aggregated volume discounts.
