Model
ImageBind
A multimodal embedding model proposed by Meta in 2023 that supports six modalities: images, text, audio, depth maps, thermal imaging, and IMU sensor signals. It achieves a unified vector space by aligning all modalities to images as a central hub, enabling cross-modal retrieval without requiring paired training data across every modality combination.
Timeline (last 90 days)
Sep 23
Meta's ImageBind model achieves a unified embedding space across six modalities.
Unverified90%
Sep 6
Representative multimodal works include OpenAI's GPT-4V, Google's Gemini, and Meta's ImageBind, with ImageBind unifying six modalities into a single embedding space.
Unverified50%