search

thien94 / ORB_SLAM2_CUDA

ORB_SLAM2 with GPU Enhancement running on NVIDIA Jetson TX1. Focus on ROS part.
Other
370 stars 119 forks source link

jetson_nano branch not OpenCV4 compatible? #21

Open MattArceri opened 3 years ago

MattArceri commented 3 years ago

The readme says to use OpenCV4 to run the jetson_nano branch on a Nano. This fails for me during compilation due to a host of issues mostly related to CvMat. As far as I can tell CvMat was entirely dropped from OpenCv4 while it was only deprecated on CV3. Is the jetson_nano branch missing some modifications to only use the new Mat class or am I missing some other workaround here?

vicent3rod commented 3 years ago

Check https://github.com/xaldyz/dataflow-orbslam/pull/2 Works on jetson nx/opencv4.

dusty-nv commented 3 years ago

Check xaldyz/dataflow-orbslam#2 Works on jetson nx/opencv4.

This doesn't appear to have the CUDA support. Here's a fork of ORB_SAM2_CUDA with OpenCV4:

https://github.com/dusty-nv/ORB_SLAM2_CUDA

RoyAmoyal commented 10 months ago

Check xaldyz/dataflow-orbslam#2 Works on jetson nx/opencv4.

This doesn't appear to have the CUDA support. Here's a fork of ORB_SAM2_CUDA with OpenCV4:

https://github.com/dusty-nv/ORB_SLAM2_CUDA

it's looks like its the same. CvMat, CvSVD etc.. its doesntlook like opencv4 support.