search

FoundationVision / GLEE

[CVPR2024 Highlight]GLEE: General Object Foundation Model for Images and Videos at Scale
https://glee-vision.github.io/
MIT License
1.02k stars 82 forks source link

跨图检测 #13

Open dawn-ech opened 5 months ago

dawn-ech commented 5 months ago

您好，我试用了GLEE，非常棒的工作！想请教一下GLEE是否支持跨图的检测呢，具体来说，就是在第一张图像上给出scribble或者bbox，然后在另一张图像上检测第一张图上的所指目标。我看到视频有类似功能，请问是否也支持静态图像呢

wjf5203 commented 5 months ago

Hi, sorry for not responding in time, and thank you for your interest in GLEE. This is a very good question. GLEE does support cross-image detection, as demonstrated by its capabilities in the VOS task—finding the same object in the second frame based on a scribble/mask/box from the first frame. Unfortunately, we have not directly tested cross-image detection in quantitative or qualitative experiments. This capability can be referenced in the VOS inference code, and we will immediately open-source this part of the code as a reference for cross-image detection. If needed, feel free to discuss further.