This was a practical programming course with a project. A number of different computer vision algorithms were implemented on the graphics card using CUDA.

• Convolution
• Diffusion
• Optical flow
• Superresolution

As part of the project, a very extensive analysis of the GPU performance using different programming techniques and different memory types was undertaken. This allowed for an optimal implementation of these algorithms.

code: convolution/diffusion

code: optical flow/superresolution

Presentation