contenido

The CVC-MUSCIMA database will be used for the competition. This database consists of 1,000 handwritten music score images, written by 50 different musicians. All the 50 writers are adult musicians in order to ensure that they have their own characteristic handwriting music style. Each writer has transcribed exactly the same 20 music pages, using the same pen and the same kind of music paper.For the competition, we will provide images without the staff lines. The staffless images are particularly useful for writer identification: since most writer identification methods remove the staff lines in the preprocessing stage, this eases the publication of results which are not dependant on the performance of the particular staff removal technique applied. Moreover, these images make easy the participation of researchers that do not work on staff removal.

For the competition the dataset is equally divided into two parts, of which the 50% of the images (500 images, 10 images from each of the writers) will be used for training the algorithms and the other 50% of the images will be used for testing them.

The test set is comprised of 500 unlabeled binary staffless images. The naming convention is test-XXX.png, where XXX ranges from 001 to 500.

The test set and its ground-truth are now available here.

The participants will be asked to send the following two files by email to afornes@cvc.uab.es:

Filename: The filename will be ParticipantCode-MethodCode.txt, where Participant Code will be the code of the participant (assigned after registration), and Method Code will be the code of the method (in this way, the participants are free to sumbit more than one method to the contest). Ex. If the participant code is CVC01 and the code of the method is MyWIdent1, then the filename will be CVC01- MyWIdent1.txt.
Content: The file must content 500 rows with the labeled writer number decided by the algorithm for each test image (e.g. the first row contains the number of the writer of the image test-001.png, the second row contains the writer number of test-002.png, etc). In this case deciding the writer is binary decision problem and it is the first nearest writer decided by the algorithm. Obviously, this number should range from 01 to 50. Thus, the output text file has the same format as the txt file provided in the training set.

A musical score will be considered as correctly classified if the writer (in this case writer number) decided by the algorithm is the same as the ground truthed one. The evaluation metric will be the classification accuracy, that is: