Code to process SK images after segmentation to get feature points
Segmentation package used is forked here: https://github.com/martindgr/image-segmentation-keras/tree/PMT_segmentation minor changes are made to give training history
train.py runs the NN segmentation training on files generated by annotate.py code here: https://github.com/martindgr/pmt-annotation
predict_multiple_from_file.py loads a model and checkpoint then predicts PMTs in images
super_module_locator is run on the segmentation output for a ring of images selected using the spreadsheet of drone information which the code also takes as input for reprojection. It can be run automatically or with manual intervention where the user decides the line lablels
adjust_number_files is used to correct any errors in this output instead of running the whole stack again, Its also useful for sorting out the half module, however take care as it only changes the super module number so always need to check the output images to make sure the inter module labelling is correct
find_bolts_and_ellipse does what it says taking in the segmentation files and super module locator output
feature_info_analysis performs cuts on ellipse fit output and produces final points files for use in photogrammetry