Researchers use AI to generate images based on people’s brain activity

Imagine if AI could turn your mental images into reality. Researchers have achieved this feat, according to an article published recently.

Researchers have found that the Stable Diffusion model of image generation can be used to reconstruct high-resolution images with high accuracy from brain activity. This was described in a December paper. They wrote that, unlike previous studies they did not need to fine-tune or train the AI models in order to create these images.

Researchers from the Graduate School of Frontier Biosciences of Osaka University said that they predicted first a latent model, which is an image’s data. They did this by analyzing fMRI signals. The model was then processed, and noise was added through diffusion. The researchers then used the decoded text from the fMRI signals in the higher visual cortex as input for a final constructed picture.

Source:
https://www.vice.com/en/article/wxje8n/researchers-use-ai-to-generate-images-based-on-peoples-brain-activity