Detects words and lines of text in a photo or screenshot (JPEG or PNG at a URL) with Amazon Rekognition, up to 100 words. Suited to signs, product labels and UI text; for documents use /v1/ocr.