About the trainning for dataset CHASE_DB1

[zllnjpi]

trafficstars

In the CHASE dataset I downloaded from the Internet, there are only images in JPG format and labels in PNG format, but no fov_ext in mask.tif format. Could you tell me how the TIF of fov_ext was obtained?

[syshin1014]

Hi, let me quote the relevant description on that from our paper. We note that: 1) FoV masks are originally not available for STARE and CHASE_DB1, 2) we used ones generated by Oliveira et al. (2018) for STARE, and 3) we have generated them for CHASE_DB1 by a simple thresholding method as in Wu et al. (2018) for comparative evaluation.