Can Labelme AI polygons be accelerated using CUDA?

wkentaro / labelme

Image Polygonal Annotation with Python (polygon, rectangle, circle, line, point and image-level flag annotation).

https://labelme.io

Other

13.52k stars 3.41k forks source link

Can Labelme AI polygons be accelerated using CUDA? #1510

Open bujiebuhuo opened 3 days ago

bujiebuhuo commented 3 days ago

If I use the SegmentAnything (Accuracy) model to obtain AI polygons now, I need to wait for many seconds. Sometimes you may want to achieve the highest possible annotation accuracy.