pple researchers have published a new paper describing their MLLM-Guided Image Editing (MGIE) AI Model, which can edit images with text prompts. Apple collaborated with University of California, Santa Barbara researchers to develop a new model capable of handling a wide range of editing scenarios, from simple color adjustments to more complex object manipulations.
The MGIE model is made up of a Multimodal Large Language Model, which expands user requests and provides “concise expressive instructions” for the diffusion model to use when editing the input image. According to the research paper, this editing method enables the MGIE model to handle “ambiguous human commands to achieve reasonable editing.”
For example, the MLLM understands a picture of a pizza with the input “make it more healthy”, which interprets the ambiguous term “healthy” and associates it with “Vegetable toppings on a pizza”. The diffusion model then edits the image as instructed by the MLLM. READ: Adopt AI, don’t sit on the sidelines: Microsoft’s Satya Nadella advises CEOs
Extant models like LLM-Guided Image Editing (LGIE) do not have the visual perception of MGIE, according to the research. With access to the input image and cross-modal understanding, the MLLM derives more descriptive instructions than the Large Language Model (LLM), which is limited to a single modality. For instance, the diffusion model will be informed about which regions to brighten by the MLLM in the MGIE model if the user wishes for the image to be brighter.
Code, data, and pre-trained models for MGIE are available for download as an open-source project on GitHub. VentureBeat claims that a web demo of the picture editing model is accessible on Hugging Face spaces. Apple has not yet disclosed how it intends to use this model outside of research initiatives.
Tim Cook, the CEO of Apple, stated earlier this month that the company is developing artificial intelligence (AI) features for its products, which will be unveiled later this year, during Apple’s quarterly earnings call. For functions like text summarization, recommendations, and more, Apple is anticipated to integrate gen-AI capabilities into its messaging app, Messages, and virtual assistant, Siri. In a similar vein, it’s likely that other Apple services like Keynotes, Pages, and Apple Music will also be subjected to AI.
Given Below are Some Adaptive Features of Apple’s |
Follow and Connect with us on
Facebook | Instagram | Linkedin | Dribbble | Twitter | Tumblr | Pinterest
