NNXRD-mfraction datasets

Datasets from "Neural networks for rapid phase quantification of Cultural Heritage X-ray powder diffraction data" needed to run the codes at: https://github.com/polinev/NNXRD-mfraction.

Some more information regarding the files:

1.  Mock-up:
   • rep-6_layer_0_0002_powder_short-iback_clean_corr_new.h5 = sample file with all XRD patterns pretreated
   • dataset_mockup_rep6.pickle = dataset for neural network training
   • best_val_loss_model_pp/h_mockup_rep6 = best model after NN training
   • mockup_rep6_rebuilt.pickle = XRD patterns rebuilt from predictions
   • REP6_seq_topas.h5 = data treated with serial Rietveld refinement
     
     
2. Historical sample:
   • S2018_157_sinogram_layer02_XRD_powder_pack_half1_short-iback_clean.h5 = data pretreated
   • S2018_157_sinogram_layer02_XRD_powder_pack_half1-back_clean_corr_mask.h5 = other sample file that contains metadata used in the code (mask and contour)
   • dataset_historical_sample_S157.pickle = dataset for NN training
   • best_val_loss_model_pp/h_article.h5 = best model after NN training
   • patterns_rebuilt_S157.pickle = XRD patterns rebuilt from predictions