Docs update and version 0.6

Change-Id: I39a122afcf58fed25e0b167c086c5429740e6210
sweeneychris · Mar 30, 2016 · ed8a8b0 · ed8a8b0
1 parent 9eb4022
commit ed8a8b0
Show file tree

Hide file tree

Showing 4 changed files with 63 additions and 46 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -92,11 +92,11 @@ set(CMAKE_RELEASE_POSTFIX "")
 set(CMAKE_DEBUG_POSTFIX "-debug")
 
 set(THEIA_VERSION_MAJOR 0)
-set(THEIA_VERSION_MINOR 5)
+set(THEIA_VERSION_MINOR 6)
 set(THEIA_VERSION_PATCH 0)
 set(THEIA_VERSION
     ${THEIA_VERSION_MAJOR}.${THEIA_VERSION_MINOR}.${THEIA_VERSION_PATCH})
-set(THEIA_ABI_VERSION 0.5.0)
+set(THEIA_ABI_VERSION 0.6.0)
 
 # THEIA data directory
 add_definitions(-DTHEIA_DATA_DIR="${CMAKE_SOURCE_DIR}/data")

diff --git a/docs/source/conf.py b/docs/source/conf.py
@@ -39,16 +39,16 @@
 
 # General information about the project.
 project = u'Theia Vision Library'
-copyright = u'2014, Chris Sweeney'
+copyright = u'2016, Chris Sweeney'
 
 # The version info for the project you're documenting, acts as replacement for
 # |version| and |release|, also used in various other places throughout the
 # built documents.
 #
 # The short X.Y version.
-version = '0.3'
+version = '0.6'
 # The full version, including alpha/beta/rc tags.
-release = '0.3.0'
+release = '0.6.0'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

diff --git a/docs/source/performance.rst b/docs/source/performance.rst
@@ -4,6 +4,8 @@
 Performance
 ===========
 
+This page was last updated March 30th, 2016.
+
 By utilizing the `Eigen <http://eigen.tuxfamily.org/dox/>`_ and `Ceres Solver
 <http://www.ceres-solver.org>`_ libraries in addition to custom scalable
 algorithms, Theia achieves state-of-the-art performance in terms of efficiency
@@ -12,8 +14,6 @@ terms of efficiency and accuracy on benchmark datasets for small and large-scale
 problems to provide meaningful context for the strengths and weaknesses of the
 library.
 
-.. note:: These results were computed using only 1 thread. Sadly, clang does not allow multi-threading with OpenMP so Ceres and other libraries cannot utilize multiple cores on the machine that the experiments were run on. We will post results from the multi-threaded experiments once they are available.
-
 
 Small Dataset Benchmarks
 ========================
@@ -28,24 +28,23 @@ reconstruction accuracy.
 
 All reconstructions were generated using the parameters found in `this
 configuration file
-<http://cs.ucsb.edu/~cmsweeney/build_strecha_reconstructions.txt>`_. Each
+<http://homes.cs.washington.edu/~csweeney/build_reconstruction_flags_strecha.txt>`_. Each
 reconstruction generated from this file was then compared to the ground truth
 reconstruction (which can be generated from the Strecha dataset using the
 "create_reconstruction_from_strecha_dataset.cc" program). The camera intrinsic
 calibration files can be easily generated from the information provided by
 datasets.
 
-================= ========== === ================= =============== ==========
-Dataset           N (input)   N  Median Error (mm) Mean Error (mm) Timing (s)
-================= ========== === ================= =============== ==========
-Castle-19         19         19  33.01             38.26           6.51
-Castle-30         30         30  29.88             32.47           11.61
-Entry-10          10         10  4.91              5.76            3.84
-Fountain-11       11         11  2.11              2.43            6.04
-Herz-Jesu-25      25         25  5.21              5.30            10.32
-Herz-Jesu-8       8          8   3.32              3.58            2.06
-================= ========== === ================= =============== ==========
+.. csv-table:: Strecha Dataset Performance
+    :header: Dataset, N (input), N, Median Error (mm), Mean Error (mm), Timing (s)
+    :stub-columns: 1
 
+    Castle-19, 19, 19, 28.2, 36.1, 2.91
+    Castle-30, 30, 30, 21.0, 33.5, 4.53
+    Entry-10, 10, 10, 4.2, 6.5, 1.76
+    Fountain-11, 11, 11, 1.6, 2.6, 2.99
+    Herz-Jesu-25, 25, 25, 4.3, 5.4, 4.53
+    Herz-Jesu-8, 8, 8, 1.9, 3.6, 1.31
 
 Large Scale Benchmarks
 ======================
@@ -60,37 +59,54 @@ meaningful reference point since incremental SfM algorithms are rather robust
 and accurate.
 
 We measure the accuracy of camera positions (approximately in meters) and timing
-results on a 2008 Mac Pro using 1 core. We report N, the number of cameras that
-could be succesfully reconstructed, in addition to the mean and median camera
-position errors after robust alignment (via RANSAC) to ground truth camera
-positions.
+results. We report N, the number of cameras that could be succesfully
+reconstructed, in addition to the mean and median camera position errors after
+robust alignment (via RANSAC) to ground truth camera positions.
 
 All reconstructions were generated using the "build_1dsfm_reconstruction.cc"
 program and the parameters found in `this config file
-<http://cs.ucsb.edu/~cmsweeney/build_1dsfm_reconstructions.txt>`_. Each
+<http://homes.cs.washington.edu/~csweeney/build_1dsfm_reconstruction_flags.txt>`_. Each
 reconstruction generated was then compared to the ground truth reconstruction
 provided in the 1dSfM dataset (these are provided as Bundler files, but can be
 converted to Theia reconstructions with the "convert_bundle_file.cc"
 program).
 
-.. note:: Results for the first three datasets will be posted soon!
-
-.. tabularcolumns:: |l|c||c|c|c|c|
-
-================= ========== ==== ================ ============== ==========
-Dataset           N (input)   N   Median Error (m) Mean Error (m) Timing (s)
-================= ========== ==== ================ ============== ==========
-Piccadilly        2152
-Union Square      789
-Roman Forum       1084
-Vienna Cathedral  836        821  4.08             13.7           2008
-Piazza del Popolo 328        325  1.27             9.2            212
-NYC Library       332        325  1.26             2.26           486
-Alamo             577        572  0.39             4.44           1060
-Metropolis        341        325  4.13             10.7           397
-Yorkminster       437        428  2.79             19.5           695
-Montreal N.D.     450        442  0.48             3.9            1624
-Tower of London   572        458  1.85             17.2           667
-Ellis Island      227        218  2.03             7.8            161
-Notre Dame        553        538  1.61             7.51           1012
-================= ========== ==== ================ ============== ==========
+.. csv-table:: 1DSfM Dataset Position Error
+    :header: Dataset, N (input), N, Median Error (m), Mean Error (m)
+    :stub-columns: 1
+
+    Alamo, 577, 561, 0.37, 1.62
+    Ellis Island, 227, 220, 1.89, 5.78
+    Madrid Metropolis, 341, 321, 1.06, 2.92
+    Montreal N.D., 450, 447, 0.42, 0.73
+    Notre Dame, 553, 547, 0.22, 0.64
+    NYC Library, 332, 321, 0.78, 6.59
+    Piazza del Popolo, 328, 328, 1.02, 4.59
+    Piccadilly, 2152, 2055, 0.72, 2.67
+    Roman Forum, 1084, 1052, 1.04, 7.75
+    Tower of London, 572, 448, 1.34, 14.12
+    Union Square, 789, 720, 3.18, 8.42
+    Vienna Cathedral, 836, 804, 1.99, 9.52
+    Yorkminster, 437, 418, 1.39, 4.90
+    Trafalgar, 5288, 4788, 4.49, 8.45
+    Gendarmenmarkt, 733, 661, 10.69, 32.65
+
+.. csv-table:: 1DSfM Dataset Timings (seconds)
+    :header: Dataset, N (input), Rotation, Position, BA, Total
+    :stub-columns: 1
+
+    Alamo, 577, 3.35, 45.22, 767.53, 874.41
+    Ellis Island, 227, 0.50, 4.59, 70.50, 94.34
+    Madrid Metropolis, 341, 1.27, 5.86, 69.62, 95.09
+    Montreal N.D., 450, 1.41, 23.87, 128.69, 207.25
+    Notre Dame, 553, 4.48, 44.38, 177.73, 372.56
+    NYC Library, 332, 0.46, 5.31, 143.60, 194.58
+    Piazza del Popolo, 328, 0.48, 8.67, 63.88, 89.19
+    Piccadilly, 2152, 49.90, 132.27, 1064.75, 1427.17
+    Roman Forum, 1084, 1.96, 23.78, 1196.35, 1302.7
+    Tower of London, 572, 0.48, 5.51, 134.72, 201.56
+    Union Square, 789, 1.16, 7.65, 73.52, 131.32
+    Vienna Cathedral, 836, 5.59, 41.85, 531.60, 764.49
+    Yorkminster, 437, 0.61, 10.06, 94.17, 164.33
+    Trafalgar, 5288, 155.72, 399.03, 544.58, 1494.19
+    Gendarmenmarkt, 733, 1.18, 14.63, 137.07, 202.261
diff --git a/docs/source/releases.rst b/docs/source/releases.rst
@@ -4,8 +4,8 @@
 Releases
 ========
 
-HEAD
-====
+`0.6.0  <https://github.com/sweeneychris/TheiaSfM/archive/v0.6.tar.gz>`_
+========================================================================
 
 New Features
 ------------
@@ -48,6 +48,7 @@ Misc.
 * Triangulation algorithms return true on success properly
 * The point cloud viewer uses the dominant plane detection to set the ground plane for viewing.
 * Upgraded EasyEXIF lib to the latest version.
+* New CHOLMOD wrapper for sparse cholesky decomposition. This provides a notable performance increase for large scale problems.
 
 `0.5.0  <https://github.com/sweeneychris/TheiaSfM/archive/v0.5.tar.gz>`_
 ========================================================================