Firefox has taken the path of using artificial intelligence features. With the upcoming release of Firefox 130 on September 3, the browser will introduce a feature that automatically creates text descriptions for images, known as the ALT tag. Nightly builds of Firefox already include this feature in the inbox PDF viewer. Initially, the option will generate descriptions only for newly added images, but the plan is to eventually extend this functionality to any image visible in the browser.
The addition of text descriptions will make screen readers more effective for people with vision problems. The AI text generator will run locally without accessing online services, similar to how the built-in translator currently works in Firefox.
The model for generating text descriptions occupies approximately 200 MB of disk space. It uses data from the DistilGPT2 model and covers 182M parameters. The browser will use a decoder based on the Vision Transformer (ViT) model to analyze images. To work with the model, the browser includes ONNX Runtume, compiled into a WASM view, and the Transformers.js library. The model loads dynamically before first use. More details here.
Support us
Winaero greatly relies on your support. You can help the site keep bringing you interesting and useful content and software by using these options:
If you like this article, please share it using the buttons below. It won't take a lot from you, but it will help us grow. Thanks for your support!
Advertisеment
How do we turn it off? Waste of processing and disk space.
Sergey, will there be ways to turn off all of this AI stuff in Windows 10 and 11 too? Maybe with Winaero Tweaker in the future?