update permissions, update README

README.md

@@ -1,7 +1,7 @@
 # 3DMatch Toolbox
 3DMatch is a ConvNet-based local geometric feature descriptor that operates on 3D data (i.e. point clouds, depth maps, meshes, etc.). This toolbox provides code to use 3DMatch for geometric registration and keypoint matching, as well as code to train 3DMatch from existing RGB-D reconstructions. This is the reference implementation of our paper:
 
-### 3DMatch: Learning Local Geometric Descriptors from 3D Reconstructions ([pdf](https://arxiv.org/pdf/1603.08182v2.pdf), [webpage](http://3dmatch.cs.princeton.edu/))
+### 3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions ([pdf](https://arxiv.org/pdf/1603.08182v2.pdf), [webpage](http://3dmatch.cs.princeton.edu/))
 
 *Andy Zeng, Shuran Song, Matthias Nießner, Matthew Fisher, Jianxiong Xiao, and Thomas Funkhouser*
 
@@ -273,7 +273,7 @@ Note: the TDF voxel grids of the scene fragments from the synthetic benchmark we
 
 See folder `3dmatch-toolbox/evaluation/model-fitting-apc`
 
-Includes code and pre-trained models to evaluate 3DMatch for model fitting on the Shelf & Tote dataset. For an evaluation example, run Matlab script `getError.m`
+Includes code and pre-trained models to evaluate 3DMatch for model fitting on the [Shelf & Tote dataset](http://www.cs.princeton.edu/~andyz/apc2016). You can download our pre-computed data (TDF voxel grid volumes for objects and scans, surface keypoints, descriptors, and pose predictions) [here](http://vision.princeton.edu/projects/2016/3DMatch/downloads/apc-intermediate-data.zip). For an evaluation example, run Matlab script `getError.m`
 
 ### Mesh Correspondence in Shape2Pose
 

depth-fusion/fuseSceneFragments.m

@@ -0,0 +1,82 @@
+% Script to fuse depth maps from real-world test scenes to generate scene 
+% fragments. Fragments are used as part of the real-world geometric
+% registration benchmark, in the same spirit as Choi et al 2015. See our
+% paper for more details.
+% 
+% ---------------------------------------------------------
+% Copyright (c) 2016, Andy Zeng
+% 
+% This file is part of the 3DMatch Toolbox and is available 
+% under the terms of the Simplified BSD License provided in 
+% LICENSE. Please retain this notice and LICENSE if you use 
+% this file (or any portion of it) in your project.
+% ---------------------------------------------------------
+
+% User configurations (change me)
+cudaBinPath = '/usr/local/cuda/bin';
+cudaLibPath = '/usr/local/cuda/lib64';
+cudnnLibPath = '/usr/local/cudnn/v5.1/lib64';
+numFramesPerFrag = 50;
+voxelSize = 0.006;
+truncMargin = voxelSize*5;
+voxelGridOrigin = [-1.5,-1.5,0.5];
+
+dataPath = '../data/test';
+sceneList = {'7-scenes-redkitchen', ...
+             'sun3d-hotel_umd-maryland_hotel3', ...
+             'sun3d-mit_76_studyroom-76-1studyroom2', ...
+             'sun3d-mit_lab_hj-lab_hj_tea_nov_2_2012_scan1_erika', ...
+             'sun3d-home_at-home_at_scan1_2013_jan_1', ...
+             'sun3d-home_md-home_md_scan9_2012_sep_30', ...
+             'sun3d-hotel_uc-scan3', ...
+             'sun3d-hotel_umd-maryland_hotel1'};
+
+% ./demo ../data/sample/depth-fusion-demo/camera-intrinsics.txt ../data/sample/depth-fusion-demo/rgbd-frames 8 8 50 -1.5 -1.5 0.5 0.006 0.03
+
+for sceneIdx = 1:length(sceneList)
+
+    sceneName = sceneList{sceneIdx};
+
+    scenePath = fullfile(dataPath,sceneName);
+    seqList = dir(fullfile(scenePath,'seq-*'));
+
+    camKPath = fullfile(scenePath,'camera-intrinsics.txt');
+
+    cloudIdx = 0;
+
+    mkdir(fullfile(fragmentsPath,sceneName));
+
+    for seqIdx = 1:min(length(seqList),3)
+
+        seqName = seqList(seqIdx).name; 
+        seqPath = fullfile(scenePath,seqName);
+        frameList = dir(fullfile(seqPath,'*.depth.png'));
+
+        % frameIdx = 0:numFramesPerFrag:(length(frameList)-1-numFramesPerFrag) % all other scenes
+        % frameIdx = 0:100:(6000-1-numFramesPerFrag) % sun3d-home_at-home_at_scan1_2013_jan_1, sun3d-home_md-home_md_scan9_2012_sep_30
+        % frameIdx = [0:100:(2750-1),7500:100:(length(frameList)-numFramesPerFrag-1)] % sun3d-hotel_uc-scan3
+        % 0:100:(length(frameList)-1-numFramesPerFrag) % sun3d-hotel_umd-maryland_hotel1
+        for frameIdx = 0:numFramesPerFrag:(length(frameList)-1-numFramesPerFrag)
+
+            % System call to fuse depth maps into a scene fragment
+            system(sprintf('export PATH=$PATH:%s',cudaBinPath));
+            system(sprintf('unset LD_LIBRARY_PATH; export LD_LIBRARY_PATH=LD_LIBRARY_PATH:%s:%s; ./demo %s %s %d %d %f %f %f %f %f %f',cudaLibPath,cudnnLibPath,camKPath,seqPath,frameIdx,frameIdx,numFramesPerFrag, ...
+                                                                                                                                                                voxelGridOrigin(1),voxelGridOrigin(2),voxelGridOrigin(3),voxelSize,truncMargin));                              
+            % Copy point cloud file to fragment folder
+            pointCloudFile = fullfile(fragmentsPath,sceneName,sprintf('cloud_bin_%d.ply',cloudIdx));
+            movefile('tsdf.ply',pointCloudFile);
+
+            % Save camera pose of base frame
+            extCam2World = dlmread(fullfile(seqPath,sprintf('frame-%06d.pose.txt',frameIdx)));
+            fid = fopen(fullfile(fragmentsPath,sceneName,sprintf('cloud_bin_%d.info.txt',cloudIdx)),'w');
+            fprintf(fid,'%s\t %s\t %d\t %d\t\n',sceneName,seqName,frameIdx,frameIdx+49);
+            fprintf(fid,'%15.8e\t %15.8e\t %15.8e\t %15.8e\t\n',extCam2World');
+            fclose(fid);
+
+            cloudIdx = cloudIdx + 1;
+        end
+
+    end
+
+end
+