search

nianticlabs / ace

[CVPR 2023 - Highlight] Accelerated Coordinate Encoding (ACE): Learning to Relocalize in Minutes using RGB and Poses
https://nianticlabs.github.io/ace
Other
360 stars 34 forks source link

Support in MacOS #8

Closed keepgoing2996 closed 1 year ago

keepgoing2996 commented 1 year ago

Is there any support to build dsacstar in macos? How can I specify opencv_inc_dir if using own python environment? Is it c++ opencv or python opencv? Thanks

tcavallari commented 1 year ago

Hello! Thanks for your question, unfortunately we don't plan to explicitly support MacOS.

Still, I'll try to answer your question to the best of my knowledge.

Is it c++ opencv or python opencv

The dsacstar bindings require the C++ version of OpenCV, but there are some parts of the Python code that require OpenCV as well (we have a few import cv2 in our scripts). The OpenCV package installed by conda (see this to create a conda environment from scratch, without using our provided environment.yml, since that's Linux-specific) provides both interfaces but, if you use a custom python environment, you'll have to make sure both the C++ libraries/includes/etc... and the cv2 Python classes are available.

One thing to note: the network training/inference code requires cuda at the moment. I know that recently PyTorch introduced a rudimental support for Metal so you could change our code to use it, but I've never tested it, and don't know how well it works (or if at all).

tcavallari commented 1 year ago

Closing for now. Happy to continue the discussion if more questions come up.