McGill 3D Shape Benchmark

Shape Analysis Group

Centre For Intelligent Machines and School of Computer Science

McGill University

Last updated 12/14/2005

 

Introduction                                      Downloads                                 Libraries and Tools

 

1. Introduction

The McGill 3D Shape Benchmark offers a repository for testing 3D shape retrieval algorithms. The emphasis is on including models with articulating parts. Thus, whereas some models have been adapted from the the Princeton Shape Benchmark and others have been downloaded from different web repositories, a substantial number have been created by us using CAD modeling tools. We provide the models in both "voxelized" form (using the .im file format, corresponding to the KUIM package) as well as in "mesh" form (using the .ply format). In the voxelized format voxels within the interior of an object have a different label than those in the background. In the mesh format each object's surface is provided in triangulated form. The mesh representations correspond to the boundary voxels of the corresponding voxel representations. They are therefore are "jagged" and may need to be smoothed for some applications. The models are organized in two databases. The first contains objects with articulating parts and the second objects whose parst undergo minor or no articulation. Within each database there are a number of ``basic'' level categories and in each there are typically between 20 to 30 examplars.

Click on the images to see the various exemplars in each category.

Objects with articulating parts

Ants Crabs Spectacles Hands Humans

Octopuses Pliers Snakes Spiders Teddy-bears

 

Objects with moderate or no part articulation

Tables Cups Chairs Airplanes Dolphins
 
Birds Four-limbs Dinosaurs Fishes  

Return to top

2. Downloads

Before downloading files, please email us so that we can know who is using this repository.


NEW! Manifold mesh versions of models below in .off format contributed by Dr. Alexander Agathos, Institute of Informatics and Telecommunications, NCSR "DEMOKRITOS", Greece. Thank you very much Alexander!


You can also download both the original ".im " format files as well as the ".ply" format mesh files. 

 

Objects with articulating parts: .including im format and .ply format (.tar.gz, 175M)

Objects with moderate or no part articulation: including .im format and .ply format (.tar.gz, 119M)

 

Small subsets:

Ants .im files .ply files Tables .im files .ply files
Crabs .im files .ply files Cups .im files ..ply files
Spectacles .im files .ply files Chairs .im files .ply files
Hands .im files .ply files Airplanes .im files .ply files
Humans .im files .ply files Dolphins .im files .ply files
octopuses .im files .ply files Birds .im files .ply files
Pliers .im files .ply files Four-limbs .im files .ply files
Snakes .im files .ply files Dinosaurs .im files .ply files
Spiders .im files .ply files Fishes .im files .ply files
Teddy-bears .im files .ply files      

Return to top

 

3. Libraries and Visualization tools

 

OpenGL should be installed on your machine.


.im files


Matlab: The .im images are raw data with a header of 1024 bytes in front. All images in the mcgill database are 128x128x128 arrays of bytes, so to read them in matlab do the following:

%open the file for reading
fid = fopen('1.im','r'); %got to the first byte after the header
status = fseek(fid,1024,'bof');
%read the data
data = fread(fid);
%close the file
fclose(fid);
%reshape the data into a 3D array
data3d = reshape(data,128,128,128);


Visualization: VTK should be installed on your machine. We have two handy applications visuim and vvisualize-s to display the .im files. These binaries were compiled for RedHat Linx and need libglut.so.3. We will make the source code available soon.

Processing package: KUIM package (http://infocom.cheonan.ac.kr/~nykwak/kuim/kuim.html)



.ply files

Visualization: There are quite a few tools online to view .ply files, such as TetView and Scanalyze.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

References

K. Siddiqi, J. Zhang, D. Macrini, A. Shokoufandeh, S. Bouix & S. Dickinson.
Retrieving Articulated 3D Models Using Medial Surfaces.
Machine Vision and Applications, 19(4), 261--274, 2008.