In this project, we take on some image processing problems with CUDA on Colab. These problems appear in CS344, an archived course on parallel programming by Nvidia and Udacity, and have been remastered in this fork to be compatible with modern OpenCV syntax and Google Colab.
In cuda-test.cu, we write a simple kernel (following one lecture) that squares every element in an array. We keep this as a clean template to CUDA's program flow.
In colab-notebook.ipynb, we discuss our full journey of CUDA with Colab, from basic experimentation, to installing nvcc4jupyter, to solving Problem 1 and Problem 2 of the CS344 problem sets. We encourage feedback for further experimentation.
To play directly on Colab, create a copy upon clicking: colab project link
A scenic view of Cinque Terre, Italy:
Conversion to greyscale:
Gaussian blurring:
