09 · Manual

Recognition and touch regions

Use text recognition and image regions as editable suggestions, then verify every label and destination yourself.

Steps

1. Open an image scene and choose Suggest text from this image when available.
2. Review Detected text before using any result as a choice label. Type the label yourself when recognition is unavailable or wrong.
3. Choose Draw this choice on the image and drag between two corners, or enter the normalized position values for precise or keyboard editing.
4. Confirm the region fits the intended visible control, has a readable text alternative, and leads to the correct scene.
5. Repeat the recipient preview after changing an image, mask, region, label, or destination.

What to expect

A person can activate the same prepared choice from the image or an equivalent labelled control.

If it does not work

• If OCR is unavailable for the language, enter labels manually. Do not change the content language just to force recognition.
• Android recognition processes selected image content on device, while Google ML Kit may separately process SDK diagnostic and usage metrics.
• If a region is invalid, keep it within the image with positive width and height, or use the numbered fields.