This is a repository of programs for deep learning object detection using Mask R-CNN. This project detects images of relict charcoal hearths (RCHs) in LiDAR images of Pennsylvania.
For additional information, please see https://zenodo.org/deposit/4593788
Geotifs images are searched for objects resembling charcoal hearths. To prepare of searching, download the files and split them into uniform sized rectangles.
Download geotifs from shared drive: All Data from Weston\LAS_Files\Part_[1-4][Area 52]QGISDEM\
Save to: /storage/images/part_[1-4]/ example: /storage/images/part_1/51blast2demPAS_Slope_DEM.tif
Run 0_1_check_that_all_DEM_Slope_tif_are_downloaded to verify all files are downloaded and to prepare to split them into rectangles.
In QGIS run qgis_4_check_we_have_all_layers.py This will check the files against a list of State Game Lands from Wikipedia. (An extra check)
12 August 2020 - This version of points is used for known charcoal hearths: Charcoal_Hearths_8-11-2020_bpc Update line 163 When this is updated, 0_split_tifs must all be run again.