cuimgops 0.1.5

CUDA-accelerated image augmentation operations for Python.

cuimgops

CUDA-accelerated image augmentation operations for Python.

Features

• Augmentation

  • Horizontal Flip
  • Vertical Flip

• Convolution

  • Gaussian Noise
  • Edge Detection
  • Blur (Smoothing)
  • Sharpening

• In Development

  • Multiple images - Batched convolution on streams

Installation

Prerequisites

• CUDA Capable GPU
• NVIDIA drivers matching your CUDA version
• CUDA Toolkit (nvcc) installed and added to PATH
• Python 3.8+ & pip

Install from PyPI

pip install cuimgops

Usage

from cuimgops import *
from PIL import Image

img = Image.open("image.jpg")

flipped = get_horizontal_flip(img)
noisy = get_gaussian_noise(img, mean=0.0, stddev=15.0)
edges = get_edge_detection(img)
blurred = get_blur(img)
sharpened = get_sharpening(img)