|
Superpixel Benchmark
Superpixel benchmark, tools and algorithms.
|
|
Superpixel Benchmark
Superpixel benchmark, tools and algorithms.
|
For getting started quickly:
For details, there are three main components that need to be built:
The following quick start was tested on Ubuntu 12.04 and Ubuntu 14.04. Make sure to clone the repository recursively, then execute:
$ git clone https://github.com/davidstutz/superpixel-benchmark --recursive $ sudo apt-get install build-essential cmake libboost-dev-all libopencv-dev cimg-dev cimg-doc cimg-examples libpng-dev libpng++-dev # Alternatively download GLog 0.3.3 to avoid the hard reset! $ git clone https://github.com/davidstutz/glog $ cd glog $ git reset --hard 0b0b022 $ ./configure $ make $ cd .. $ cd superpixel-benchmark $ mkdir build $ cd build # Better install GLog properly and adapt cmake/FindGlog.cmake! $ cmake .. -DGLOG_ROOT_DIR=~/glog $ make
Installs Cmake, Boost, OpenCV, GLog, libpng, libpng++ and CImg. This builts the seven recommended algorithms as discussed in the paper as well as the benchmark and tools.
Also see .travis.yml for details on building on a fresh Ubuntu 14.04 installation!
Building the C++ components is done using CMake and all of them are based on the lib_eval component which holds the benchmark as well as toopls.
A C++ compiler supporting C++11 is assumed to be available. It was tested with gcc >= 4.8.4. Building has been tested with Ubuntu 12.04 and Ubuntu 14.04.
All algorithms depend on the tools in lib_eval. Requirements are:
Additionally, the algorithms built by default depend on:
Note that the required CMake modules, e.g. for finding GLog, can be found in cmake in case these need to be adapted.
CMake, OpenCV and Boost can be installed as follows:
$ sudo apt-get install build-essential cmake libboost-dev-all libopencv-dev
OpenCV can alternatively be installed following these instructions. Currently, OpenCV 2.4.11 is supported, in general 2.4.x should work fine. For upgrading to OpenCV 3.0 and OpenCV 3.1, it might be necessary to update constants (e.g. as used for color conversion). Some implementations are known to work with OpenCV 3 and OpenCV 3.1 (e.g. reSEEDS).
GLog 0.3.3 should be installed manually. GLog should be downloaded or cloned from google/glog. If the repository is cloned, make sure to checkout version 0.3.3. Then:
$ cd glog-0.3.3/ $ ./configure $ make $ sudo make install
For further instructions see the issue tracker at google/glog.
Note that GLog can alternatively be installed using
$ sudo apt-get install libgoogle-glog-dev
However, cmake/FindGlog.cmake needs to be adapted and some parts might not working with newer versions.
For installing PNG and PNG++:
sudo apt-get install libpng-dev # should already be installed for OpenCV sudo apt-get install libpng++-dev
As reference, these are the library versions as installed on Ubuntu 14.04 (checked using dpkg -l) where OpenCV and GLog where installed manually:
||/ Name Version Architecture +++-===================-==============-============== ii gcc 4:4.8.2-1ubunt amd64 ii cmake 3.2.2-2ubuntu2 amd64 ii libboost-dev 1.54.0.1ubuntu amd64 ii libpng12-dev 1.2.50-1ubuntu amd64 ii libpng++-dev 0.2.5-1 all
After verifying that the requirements are met:
$ mkdir build $ cd build $ cmake .. $ cmake -LAH
This will list all available CMake options. These options include:
-DBUILD_CCS: build CCS (Off)-DBUILD_CIS: build CIS (Off), follow building CIS for details-DBUILD_CRS: build CRS (On)-DBUILD_CW: build CW (Off)-DBUILD_DASP: build DASP (Off)-DBUILD_ERGC: build ERGC (On)-DBUILD_ERS: build ERS (On)-DBUILD_ETPS: build ETPS (On)-DBUILD_FH: build FH (Off)-DBUILD_EXAMPLES: build examples (Off)-DBUILD_LSC: build LSC (Off)-DBUILD_MSS: build MSS (Off)-DBUILD_PB: build PB (On)-DBUILD_PRESLIC: build SLIC (Off)-DBUILD_REFH: build reFH (Off)-DBUILD_RESEEDS: build reSEEDS (On)-DBUILD_SEEDS: build SEEDS (On)-DBUILD_SLIC: build SLIC (On)-DBUILD_VC: build VC (Off)-DBUILD_VCCS: build VCCS (Off)-DBUILD_VLSLIC: build vlSLIC (Off)-DBUILD_W: build W (Off)Note that the algorithms recommended in the paper are built by default. To change this, use the options indicated above or edit CMakeLists.txt accordingly.
In the following, we outline additional prerequisites needed to build selected superpixel algorithms. Note that when using the default options, no additional requirements need to be fulfilled.
The CIS source code is not directly included in this repository due to license constraints. For installing and building CIS, follow these instructions.
Eigen3 is required:
sudo apt-get install libeigen3-dev
For reference, the following version was installed on Ubuntu 14.04:
||/ Name Version Architecture +++-===================-==============-============== ii libeigen3-dev 3.2.0-8 all
sudo apt-get install cimg-dev cimg-doc cimg-examples
For reference, the following version was installed on Ubuntu 14.04:
||/ Name Version Architecture +++-===================-==============-============== ii cimg-dev 1.5.7+dfsg-1 all
Note that the source code in lib_vlslic is part of the VLFeat library. The library was stripped to the essentials components needed for vlSLIC. Alternatively, VLFeat can be installed as described here. The corresponding CMakeLists.txt in vlslic_cli and lib_vlslic may need to be adapted (note that lib_vlslic/vlslic_opencv.h is header only such that lib_vlslic might not be required to be compiled.
Some MatLab sources are based on C++ components. In order to compile and MEX these components, make.m can be used. It provides the following variables:
BUILD_TOOLS = 1; BUILD_EAMS = 0; BUILD_NC = 0; BUILD_QS = 0; BUILD_SEAW = 0; BUILD_TP = 0; BUILT_TPS = 0;
These can be used to compile individual algorithms. Alternatively, the corresponding make.m files can be run directly, i.e. lib_tools/make.m, lib_eams/make.m, lib_nc/make.m, lib_qs/make.m, lib_seaw/make.m, lib_tp/make.m and lib_tps/make.m.
Compiling was tested with MatLab R2015b and gcc >=4.8.4. Note however, that later MatLab version have different requirements regarding the compiler. For example, in MatLab R2015b, gcc 4.8.4 is not officially supported (still it works in this case).
Only PF is written in Java. For compiling the Java source, follow lib_pf/make.sh:
$JAVAC *.java $JAR cfm PathFinder.jar Manifest.txt *.java *.class $JAVA -jar PathFinder.jar
Make sure to set the variables JAVAC, JAR and JAVA correctly. On Ubuntu 14.04, this may look as follows:
JAVAC="/home/david/jdk1.8.0_45/bin/javac" JAR="/home/david/jdk1.8.0_45/bin/jar" JAVA="/home/david/jdk1.8.0_45/bin/java"
1.8.11