When an image is scanned by Panorama, we want to ensure everyone, including those using screen readers, can understand it. While Panorama allows manual entry of alternative text for images, content creators can generate descriptions automatically using its AI-powered system. This article provides a simple breakdown of how Panorama generates alternative text.
Privacy Assurance
We respect the privacy of individuals and do not use your images or any customer data to train our AI service. The AI model is trained on a separate data set and does not learn from your specific images.
How Panorama Generates Alternative Text
-
Image Extraction
If the image is located within a document, it is first extracted to isolate it from additional content. -
AI Analysis
Once the image is extracted, it is sent to our AI service. This service has been trained on millions of images and can recognize a wide variety of objects, scenes, and activities. -
Label Detection
Our AI service then analyzes key elements within the image, providing labels that best describe the image. For example, the AI can detect objects like "cats," "bicycles," and "trees," and activities like "running" and "swimming." -
Generating Alternative Text
Using the labels provided by the AI, our system generates descriptive text for the image. This text, known as "alternative text," is a brief description that conveys the content and context of the image. -
Applying Alternative Text
Finally, the generated alternative text is added to the image, ensuring that users who use assistive technologies such as screen readers can understand it.