Saliency-object-detection

Dependency

CMAKE 3.11.3 OPENCV 3.5.1_5

For Mac

brew install cmake
brew install opencv

Directory

.  
├── main      
    ├── dip.cpp                   # our main source code    
    ├── CMakeList.txt             # cmake configuration, which includes some options to print processing image   
    ├── compile.sh                # compile   
    ├── build.sh                  # cmake
    ├── segementation             # segementation algorithm implemented by davidstutz[1]
├── include                       # implement region based saliency    
    ├── saliencyMap.hpp                
    ├── saliencyMap.cpp    
├── test_img                      # Some test image    
├── result_img                    # Result from test image   

├── Global Contrast Based Salient Region Detection.pdf   # Our slides in pdf format    
├── Global Contrast Based Salient Region Detection.pptx  # Our slides in pptx format    
├── Team4_ProposalReport.pdf                             # Our proposal    
└── README.md

[1] davidstutz's segmentation implementation

Usage

All operations are under main/

cd main/

Compile

sh build.sh
sh compile.sh

Execute

./DIP [Path/To/Image]

Play with parameter

vim ../saliencyMap.hpp
sh compile.sh
./DIP [Path/To/Image]

All parameters are specified in Global Contrast Based Salient Region Detection.pdf

Options

There are some options for showing addtional image during the process.

vim CMakeLists.txt /* uncomment your option */

alt

To be fix

Some of the images may lead to segmentation fault due to our function find_largest_connect_region can not handle the cases when the connected region grows too large.