Hello Aaron, Thank you for your tutorial, this helped me a lot! One thing I would like to add: When exporting the training data for deep learning, class names (labels) MUST be different from 0. For example, if you are using Mask R-CNN to delineate objects and to label different classes (e.g. vegetation, buildings, etc.), start class names at 1, not 0.
@keesillem5810 Жыл бұрын
Hi! This is a very helpful video! For my thesis, I am exploring different deep-learning methods for object detection in Lidar-derived data. Mask R-CNN is one of them. I am trying to export my data (as you did) using the "export data for deep learning tool", somehow I can only export a TIFF file instead of PNG. I am not sure if this matters for the rest of the project but if I try to run the Train deep learning model tool, I get an error that says No module named "torch". I am not sure if I downloaded pytorch correctly. Where do you think the problem lies? Thank you!
@ColumbiaBasinJim2 жыл бұрын
I apologize if this is a novice question but why download LiDAR point clouds to build DEMs if LiDAR-derived DEMs are also available for downloading?