How to pick specific file name in directory

Greeting!

I Have a problem about getting bunch of input picture with specific name. My input consist of

crop_001_S_01
crop_001_S_02
crop_001_S_03
.....
crop_002_S_01
crop_002_S_02
crop_003_S_03
.....

Where (001) represent the single individu and (S_01) representing 1 picture taken from single person. There is 10 picture for single person set ( crop_001_S_01 - crop_001_S_10). There is more than 800++ picture in single folder.

What I want to do is to take only specific number of picture from single person as train image like I only take second until forth picture from every person

crop_002_S_02 - crop_002_S_04
crop_003_S_02 - crop_003_S_04
crop_004_S_02 - crop_004_S_04
crop_005_S_02 - crop_005_S_04
......

I try to write like below

in_name = 'S_0';

for fig = 2:3

     testdir = dir(['C:\FYP\work\FYP\rawpic\*', in_name(fig),'*.jpg']);

end

Yet, I still could not get the specific file required. It's there a way to pick a specific input picture name from dir? Thanks!

[dpb]

dir wildcards aren't sufficiently powerful to do the specific selection you're looking for; however you can return the full directory listing and use regular expressions on the .name array to select any subset desired--just have to be able to write a match expression.

The alternative is to read the directory names and break apart into the fields and do logical operations on them; also doable, which flavor to choose is your choice.

[Stephen23]

dir does not offer this kind of control. But regular expressions do: you could easily read all of the filenames into a cell array, and then filter out the ones that you want:

S = dir(...);
N = {S.name};
... filter here

[Muhammad Bariq Azmi]

So, I need to screen the name using name directory? Like file.name{i}?

[dpb]

Not exactly what we're suggesting, no...as Stephen wrote, first return the DIR() output for the group of interest or all the files and just only worry about screening from the full pool--that's the struct S in above code snippet.

Then cast the names field of the structure into a cell array; now you have a searchable cellstr array of all the names from which to do the selection matching however you wish; a regexp expression can undoubtedly be written automagically to meet whatever subset it is you wish out of those.

Answer 1

[Akira Agata]

As Stephen-san and dpb-san mentioned, regular expression will help extract files based on file name. Here is one example to extract the second until fo[u]rth picture from every person.

fileList = struct2table(dir('*.jpg'));

match = regexp(fileList.name,'crop_[0-9]{3}_S_(02|03|04).jpg');
idx = cellfun(@isempty,match);
fileList.isTrain = ~idx;

[Muhammad Bariq Azmi]

Thank you, I already got the file and filter to only image input that I want. Now I can proceed with these input. Thank you again!

[dpb]

And may the forth be with you... :)