GPU accelerated edge detection

For this project you need the following dependencies:

• CUDA 9.1 or higher
• python3, numpy, matplotlib, mpl_toolkits which can be easily installed through pip3
• an NVIDIA GPU

To see the test we performed for the project you should type:

• $ bash run_tests.sh
• if you are interested in the raw data used for the plots in testSpeedConv.cu, please comment the lines 49 and 50 of run_tests.sh. They will be stored in ./tests/speed_conv_img_size_CPU.txt and ./tests/speed_conv_img_size_GPU.txt. After inspecting them run $ bash clean.sh to remove unnecessary files.

To use an application of the project you should type. You can the use the 1.jpg image as an example:

• $ bash app.sh path_to_a_square_image

To get the results of the application with the best parallelization method in our 15 selected figures you should type (the result will be stored in figures/resulting_images):

• $ bash doAllImage.sh

To clean your directory of unnecessary text files or resultng images you should type:

• $ bash clean.sh

The report of the project can be found in:

• report