Takes two input images; one as a subject (with a green screen background) and another as a background, and creates an output image with the former's foreground placed on the provided background.
This is available using AVX2 intrinsics in the GreenScreen_AVX folder and can be compiled using the -mavx2 flag.
A CUDA implementation is also available in the GreenScreen_CUDA folder which can be run in Google Colab (note that a GPU accelerator must be selected for your workspace).
Also, a sharpening algorithm has been implemented in both versions which can slightly sharpen the output.
Inverts an image, implemented as a CUDA warmup.
An implementation of an algorithm sped up using CUDA.
An edge-detection implemented using a simple Sobel filter using CUDA. Works with high-res (8K) images, and includes brightening and threshold filters. A single-thread implementation is provided to demonstrate speedup.