DLV Intellifeatures

Linux/Mac Installation Instructions

1. Install Python 2.7 and pip using the instructions found here: https://wiki.python.org/moin/BeginnersGuide/Download

2. Install virtualenv using pip

       pip install virtualenv
   

3. Clone the DLV repository using the following command:

       git clone https://github.com/joshua-smith4/DLV_intellifeatures.git <path_to_DLV>
   

4. Create a python virtual environment in the DLV directory

       cd <path_to_DLV>
       virtualenv -p <python2.7_executable> .
       source bin/activate
   

5. Download and install opencv with python bindings using the instructions found here (make sure to build python bindings):

       https://docs.opencv.org/3.4.5/d7/d9f/tutorial_linux_install.html
       Make sure to include configuration option PYTHON2_INCLUDE_DIR2 as <path_to_DLV>/include/python2.7
   

6. Download and install z3 as shown here: https://github.com/Z3Prover/z3 (Remember to follow the instructions for Python install with virtualenv)

       git clone https://github.com/Z3Prover/z3.git <path_to_z3>
       cd <path_to_z3>
       source <path_to_DLV>/bin/activate
       python scripts/mk_make.py --python
       cd build
       make -j4
       make install
   

7. Install all other dependencies with:

       pip install -r requirements.txt
   

8. Set up keras environment with a keras config file at ~/.keras/keras.json with the text:

       {
           "floatx": "float32",
           "epsilon": 1e-07,
           "backend": "theano",
           "image_data_format": "channels_last",
           "image_dim_ordering": "th"
       }
   

9. To run the DLV, checkout the desired branch

       git checkout <desired_branch> (eg. origin/orig_dlv, origin/intellifeatures)
       python main.py --dataset=[mnist,cifar10,gtsrb] --mode=[train,read] --numtestimages=N --numfeaturedims=M
   

10. In order to run GTSRB, you must first download the training dataset here: http://benchmark.ini.rub.de/Dataset/GTSRB_Final_Training_Images.zip and place the folder "Final_Training" at the location networks/GTSRB

11. For faster run time by connecting to a supported GPU follow the instructions here: https://keras.io/getting-started/faq/#how-can-i-run-keras-on-gpu

Windows Installation Instructions

1. Download and install git by following the instructions here: https://git-scm.com/downloads

2. Download and install the Visual Studio C++ build tools using visual studio installer

3. Install Python 2.7 and pip using the instructions found here: https://wiki.python.org/moin/BeginnersGuide/Download

4. Open a Visual Studio Developer Command Prompt

5. Install virtualenv using pip

       pip install virtualenv
   

6. Clone the DLV repository using the following command:

       git clone https://github.com/joshua-smith4/DLV_intellifeatures.git <path_to_DLV>
   

7. Create a python virtual environment in the DLV directory

       cd <path_to_DLV>
       virtualenv -p <python2.7_executable> .
       Scripts\\activate.bat
   

8. Download and install opencv using the instructions found here (make sure to build python bindings):

       https://docs.opencv.org/3.4.5/d3/d52/tutorial_windows_install.html
       (https://sourceforge.net/projects/opencvlibrary/files/opencv-win/3.4.3/opencv-3.4.3-vc14_vc15.exe/download)
   

9. Download and install z3 as shown here: https://github.com/Z3Prover/z3 (Remember to follow the instructions for Python install with virtualenv)

       git clone https://github.com/Z3Prover/z3.git <path_to_z3>
       cd <path_to_z3>
       <path_to_DLV>\\Scripts\\activate.bat
       python scripts/mk_make.py --python
       cd <path_to_z3>\\build
       nmake
   

10. Add <path_to_z3>\build\python to the PYTHONPATH environment variable and <path_to_z3>\build to the PATH environment variable.

11. Install all other dependencies with:

       pip install -r requirements.txt
    

12. Set up keras environment with a keras config file (keras.json generally found at C:\Users\) with the text:

       {
           "floatx": "float32",
           "epsilon": 1e-07,
           "backend": "theano",
           "image_data_format": "channels_last",
           "image_dim_ordering": "th"
       }
    

13. To run the DLV, checkout the desired branch

       git checkout <desired_branch> (eg. origin/orig_dlv, origin/intellifeatures)
       python main.py --dataset [mnist,cifar10,gtsrb,imageNet]