Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Alt text, or alternative text, is an HTML attribute added to an <img>
tag that displays images on an application or web page. It looks like this in plain HTML code:
<img src="elephant.jpg" alt="An elephant in a grassland.">
Alt text enables website owners to describe an image in plain text. These image descriptions improve accessibility by enabling screen readers such as Microsoft Narrator, JAWS, and NVDA to accurately communicate image content to their users who are blind or with visual impairment.
Alt text is also important for image search engine optimization (SEO). It helps search engines understand the visual content in your images. The search engine is then better able to include and rank your website in search results when users search for the content in your website.
Image Analysis offers image captioning models that generate one-sentence descriptions of an image's visual content. You can use these AI-generated captions as alt text for your images.
Auto-generated caption: "An elephant in a grassland."
Microsoft’s own products such as PowerPoint, Word, and Edge browser use image captioning by Image Analysis to generate alt text.
- Improve accessibility and user experience for blind and low-vision users. Alt Text makes visual information in images available to screen readers used by blind and low-vision users.
- Meet legal compliance requirements. Some websites may be legally required to remove all accessibility barriers. Using alt text for accessibility helps website owners minimize risk of legal action now and in the future.
- Make your website more discoverable and searchable. Image alt text helps search engine crawlers find images on your website more easily and rank them higher in search results.
To ensure accurate alt text for all images, you can choose to only accept captions above a certain confidence level. The right confidence level varies for each user depending on the type of images and usage scenario.
In general, we advise a confidence threshold of 0.4
for the Image Analysis 3.2 API.
On rare occasions, image captions can contain embarrassing errors, such as labeling a male-identifying person as "woman" or labeling an adult woman as "girl". We encourage users to consider using the latest Image Analysis 4.0 API which eliminates some errors by supporting gender-neutral captions.
Please report any embarrassing or offensive captions by going to the Azure portal and navigating to the Feedback button in the top right.
Image captions are available in English, Chinese, Portuguese, Japanese, and Spanish in Image Analysis 3.2 API. In the Image Analysis 4.0 API, image captions are only available in English.