about image ids in posts.txt of mediaeval2016 devset

[MeHibird]

Hello, the image ids are sandy_real(fake)_XX In Mediaeval2016 devset posts.txt, but the image files I found in mediaeval2015 dev+test set are named with sandyA_real(fake)XX or sandyB_real(fake)XX. And the sandyA and sandyB images are different ,so I don't know which one should I keep and use.
And I cannot find images named with sandy_real(fake)_XX.

[yingqichao]

I encountered the same problem

[yingqichao]

@MeHibird If you have solved the problem please kindly help me solving this. Many many thanks! Mail: [email protected]

[olgapapa]

Dear @MeHibird and @yingqichao,

For the correct correspondence of the image filenames, you can refer to the file set_images.txt.

The datasets were created as parts of the MediaEval2015 and 2016 Verifying Multimedia Use task. For the MediaEval2016, we shared as devset the dev + test sets of MediaEval2015. Additionally, we shared an original dataset for MediaEvale2016 testset.

Thus, the images in posts.txt are those provided in MediaEval2015 and specifically in the folders MediaEval2015 devset and MediaEval2015 testset.

An example, in set_images.txt the filename of the image contained in the tweet 263110236175466497 is sandyA_fake_03 and in posts.txt the same tweet’s image filename is sandy_fake_17. Using the tweet IDs you can find the corresponding names for all images.