Data sets: Deep Neural Net Instance Segmentation of Optical Floating Zone Melts

N. Carey, W.A. Phelan, A. Rachidi, C. Krill, P. Appel, J. Zahn, M. Hudes, B. Schuster, T.M. McQueen, and D. Elbert

Image and JSON labeled polygon training and test sets for deep-learning models. GoodMelts, fastBottom, and fastTop archives contain dataset for each of the named instance classes described in the associated paper (DOI pending). 10x_crossVal archive contains the training data split into the 10 k-fold, cross validation sets used in the study. TestSet are a reserved set of labeled images used for model evaluation, but not used in training.

ItemLink
goodMeltsPARADIM_BC_LDFZ_goodMelt.zip
fastBottomPARADIM_BC_LDFZ_fastBottom.zip
fastTopPARADIM_BC_LDFZ_fastTop.zip
10x crossValPARADIM_BC_LDFZ_10_fold_crossVal.zip
testSetPARADIM_BC_LDFZ_testSet.zip