Spaces:

haotiz

/

glip-zeroshot-demo

like 70

Build error

GLIP (Grounded Language-Image Pre-training) is a generalizable object detection (we use object detection as the representative of localization tasks) model. It is language aware, taking a natural language prompt as instruction. It is also semantic rich, able to detect millions of visual concepts out-of-box. GLIPv2 further extends such ability to instance segmentation and grounded vision-language understanding tasks; see examples in Figure 2. GLIP introduces language into object detection and leverages self-training techniques to pre-train on scalable and semantic-rich data: grounded image-captions (24M). This marks a milestone towards generalizable localization models GLIP enjoys superior zero-shot and few-shot transfer ability similar to that of CLIP/GPT-2/GPT-3.

Github: https://github.com/microsoft/GLIP
Hugging Face Demo: https://huggingface.co/spaces/haotiz/glip-zeroshot-demo
CVPR 2022 Best Paper Finalist (1 out of 33)
Microsoft Research Blog: https://www.microsoft.com/en-us/research/project/project-florence-vl/articles/object-detection-in-the-wild-via-grounded-language-image-pre-training/
Twitter: https://twitter.com/HaotianZhang4AI/status/1569775843681128448

Please consider granting free T4 access to this project. We hope the open-source and radio demo will further arouse the interests in multimodal intelligence and vision-language research.