嵌入
空格(标点符号)
计算机科学
人工智能
操作系统
作者
Rohit Girdhar,Alaaeldin El-Nouby,Zhuang Liu,Mannat Singh,Kalyan Vasudev Alwala,Armand Joulin,Ishan Misra
标识
DOI:10.1109/cvpr52729.2023.01457
摘要
We present ImageBind, an approach to learn a joint embedding across six different modalities - images, text, audio, depth, thermal, and IMU data. We show that all combinations of paired data are not necessary to train such a joint embedding, and only image-paired data is sufficient to bind the modalities together. ImageBind can leverage recent large scale vision-language models, and extends their zero-shot capabilities to new modalities just by using their natural pairing with images. It enables novel emergent applications 'out-of-the-box' including cross-modal retrieval, composing modalities with arithmetic, cross-modal detection and generation. The emergent capabilities improve with the strength of the image encoder and we set a new state-of-the-art on emergent zero-shot recognition tasks across modalities, outperforming specialist supervised models. Finally, we show strong few-shot recognition results outperforming prior work, and that ImageBind serves as a new way to evaluate vision models for visual and non-visual tasks.
科研通智能强力驱动
Strongly Powered by AbleSci AI