Home Technology Meta’s ImageBind AI, an open-source project, seeks to simulate human perception.

Meta’s ImageBind AI, an open-source project, seeks to simulate human perception.

Meta is launching a new AI tool called ImageBind, which is capable of identifying connections between data in a way that resembles human perception. Unlike other image generators such as Midjourney, Stable Diffusion and DALL-E 2, which pair words with images and create visual scenes from text descriptions, ImageBind is able to connect text, images/videos, audio, 3D measurements, temperature data and motion data together, without the need to train on every possibility beforehand. Over time, this has the potential to enable more complex environments to be generated from basic inputs such as a text prompt, image or audio recording.

This technology brings machine learning closer to human learning by mimicking the way our brains process sensory experiences to derive information about our environment. As computers get closer to replicating multi-sensory connections, tools like ImageBind have the capacity to generate fully sensory scenes based on limited data chunks.

Meta’s developers anticipate ImageBind opening up new opportunities to create animations by combining images with audio prompts. It could enable immersive videos to be created with realistic soundscapes and movement based on just text, image or audio input. This technology could also be beneficial in the accessibility sphere, where real-time multimedia descriptions could help visually or hearing-impaired individuals to better understand their surroundings.

Meta’s ambition for the technology hints at the wider potential of VR, mixed reality and the metaverse – for example, a headset that can generate fully realized 3D scenes with sound and movement on-the-fly, or virtual game developers using it to simplify the design process. While ImageBind only explores six “senses” at present, Meta’s developers foresee it being used to develop richer AI models that covers senses such as touch, speech, smell and brain fMRI signals in the future. To gain access to this sandbox, developers are invited to dive into Meta’s open-source code, which is available on Github.

Meta’s graph showing ImageBind’s accuracy outperforming single-mode models.

Meta

 

Reference

Denial of responsibility! TechCodex is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessory action within 24 hours.

Denial of responsibility! TechCodex is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
DMCA compliant image

Leave a Comment