[indian] Adding missing sample
Fix #47 (closed). The sample 'DatasetA/Training/nx23.jpg' was added to the 'default' protocol and to each of the 10 folds of the 'indian' dataset. This is the only missing sample and consequently the only addition to the protocols.
Merge request reports
Activity
added New Feature label
assigned to @ojimenez
Thanks, @ojimenez - could you explain the reasoning of the subsets where you inserted the sample at? For example, why inserting the sample at the
validation
set on thedefault
evaluation protocol instead oftrain
? (N.B.: There is no right answer - it is just for keeping the record of decisions somewhere!)Edited by André Anjos@andre.anjos I noticed that the subsets and the folds are balanced in the number of negatives and positives. However, looking into the
default
split more carefully, the missing sample fits better in thetrain
set. I have now updated this in thedefault
protocol and added the missing sample to the train set. The resultingdefault
protocol now has the same amount of positive and negative cases in all of the subsets.Regarding the folds, I added the missing sample (DatasetA/Training/nx23.jpg) in the test set of
fold-0
, in the train set in half of the folds, i.e. 1,3,5,7,9, and in the validation set in the remaining folds, i.e. 2,4,6,8Edited by Oscar JIMENEZI relaunched the synchronisation on both the Linux and macOS hosts. The following files were uploaded:
DatasetA/Training/nx23.jpg DatasetB/Testing/nd152.dcm DatasetB/Testing/nd153.dcm DatasetB/Testing/nd169.dcm DatasetB/Testing/nd234.dcm DatasetB/Testing/nd436.dcm DatasetB/Testing/nd858.dcm DatasetB/Testing/pd149.dcm
Thanks, @ojimenez - looks merry!
mentioned in commit 2ed2f09a