SHREC'2015: Scalability of Non-rigid Shape Retrieval
====================================================

NOTE: If you use this information for your research, please cite the article:

 -  Sipiran, I.; Bustos, B.; Schreck, T.; Bronstein, A. M.; Bronstein, M.; Castellani, U.; Choi, S.; Lai, L.; Li, H.; Litman, R.; Sun, L.: 	"Scalability of Non-Rigid 3D Shape Retrieval". Proc. Eurographics Workshop on 3D Object Retrieval (3DOR). Pages 121-128. 2015.


The package contains the information needed to reproduce the results of the paper. 

For the precision-recall plot, please use the file "precision_recall2.dat". The columns are organized as follows:
  - Column 1: Recall values
  - Column 2: Method SQFD-HKS
  - Column 3: Method SQFD-SIHKS
  - Column 4: Method Sparse-BoF
  - Column 5: Method GDD-Lp
  - Column 6: Method GDD-Corr
  - Column 7: Method GA

For the bar plot (Fig.1 in the paper), the file "result.dat" contains the relevant data.

To groundtruth of the benchmark is contained in files "target.cla" and "query.cla". This files should be used along with the provided evaluation application.  

In order to compile the evaluation code, we provide a CMake project. The two steps to compile are:
  - Run CMake in the source folder (it will generate a Makefile): 
	>> cmake .
  - Compile (it will produce a binary "evaluation_shrec2015"):
	>> make

Finally, to execute the evaluation, the command is
	>> evaluation_shrec2015 target.cla query.cla your_distance_matrix -bin value -all
where,
	- your_distance_matrix is a text file which stores the 229 x 96,486 distance matrix.
	- value is the number of recall values, tipically 10

