Meta open-sources ImageBind, a multi-modal AI model that combines text, sound and visual data

Unlike OpenAI and Google’s AI research, which is increasingly closed, Meta AI still maintains an open attitude. On Tuesday (5/9), the multi-modal AI model ImageBind was open sourced, and ImageBind is an AI system that integrates 6 major perception forms. , including photo/video, text, sound, depth, heat and inertial measurement unit (IMU).