This repo provides a demo to get dense correspondences from images for beginners, depending on the libraries of Detectron2 and DensePose, thanks.
Input Image | CSE |
---|---|
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
Note: These images are derived from DensPose LVIS.
This code works well on MacOS and Linux, after following settings:
-
install torch >= 1.9.0
-
install Detectron2 (See Details)
-
install DensePose:
pip install git+https://github.com/facebookresearch/detectron2@main#subdirectory=projects/DensePose
Note: You may uninstall cv2 before installing Detectron2, to avoid some erros.
You can get more examples from Detectron2.