Rank list format for SHREC 2013 - Large Scale Sketch-Based 3D Shape Retrieval

The participants should submit one separate rank list file for each sketch in the query set. The file must be an ASCII file and the filename should be the same as the query model.

The first line should contain the name of the query sketch.

Each of the following lines should contain the name of the retrieved target model and its distance to the query sketch in the order of decreasing similarity. The name and the distance should be separated by white space.

For the track "Large Scale Sketch-Based 3D Shape Retrieval", there are 7200 hand-drawn query sketches, so the participants should totally submit 7200 (4500 for the training dataset and 2700 for the testing dataset) separate rank-list files in the required format. The files should be collected in a single archive file and the name of the archive file should indicate the retrieval method. The archive file should be sent via e-mail Bo Li. If the participants have more than one method, they should submit an archive file per method containing the 4500 (for the training dataset) or 2700 (for the testing dataset) rank-list files (the filename should indicate the method).

An example rank list file can be downloaded here: 1921

Example rank list format for the query sketch "1921.png":
The filename of the rank list should be 1921 with no file extension.
The content of the rank list should be as follows:

m665 19.874407326983
m64 20.007586885119
m626 20.296906356373
m791 20.354548230224
m6 20.373317409476
m797 20.741283192480
m629 20.812271004532
....... .......

Example archive file for the method X should have a name method_X.zip or method_X.rar. The file should contain 4500 (for the training dataset) or 2700 (for the testing dataset) rank lists of the query sketches.