LBP Descriptor CUDA

Project for the Parallel Computing exam at the University of Florence.

The sequential and first parallel implementations for the same project can be found here: LBP descriptor.

Local binary patterns (LBP) is a type of visual descriptor used for classification in computer vision. It has since been found to be a powerful feature for texture classification.

Directories Layout

├── src                       # Source files
│   ├── ...
├── res                       # Resources files
│   ├── histograms            # Histograms csv and plot
│   ├── images                # Images 
│   │   ├── ppm               # Images for experiments in ppm format
│   ├── plots                 # Plots
│   ├── results               # Experiments results
├── docs                      # Documentation files
│   ├── LBP_descriptor_paper  # LBP analysis and experiments results 
│   ├── LBP_descriptor_slides # LBP presentation of LBP implementation and experiments results 

Dataset

Images used for experiments are in PPM format and some examples can be found in res/images/ppm folder.

Run experiments

To run CUDA experiments, follow the commands below:

Compile files:

nvcc -c src/main.cu src/PPM.cpp src/Image.cpp

Link files:

nvcc Image.o PPM.o main.o -o main

And run executable:

./main

Authors

This project was carried out in collaboration with Francesca Del Lungo for the Parallel Computing exam.