Cleaning the groundtruth images from the LowRes_13434 dataset

The ground truth images from the LowRes_13434_overlapping_pairs.h5 dataset contains spurious pixels.This dataset is fixed:

Initial dataset with spurious pixels (Top). Example of cleaned ground truth image (Down)

The cleaned dataset is now available on github as a compressed hdf5 file.

The images were cleaned as follow:

Generation of 82146 chromosomal overlappings from pairs of chromosomes

The following jupyter notebook was published on github . The aim is to generate a large dataset of overlapping chromosomes (grey scaled image + ground truth label image) to train a neural network to perform semantic segmentation on such images.
To gain a large number of images the resolution was decreased by 16. A first try proposed in the ai.on project seems to do a very good job. The results was obtained from a dataset of 13434 pair of images from a python implementation of Unet.