img-preprocess-pipeline 0.0.2

Customized library for preprocessing images

Library: img-preprocess-pipeline

📋 Table of Contents

1. 📖 Project description
2. 🔨 Installation
3. ⚙️ Functions
4. ☕ Author and contact
5. ©️ License

📖 Project description

The img-preprocess-pipeline package provides a comprehensive solution for loading, displaying, and preprocessing images, streamlining their use in machine learning projects. Its features include:

• Image Loading: Import images from various sources and formats.
• Image Displaying: Quickly and intuitively visualize images.
• Image Preprocessing: Prepare images for use in machine learning models through:
  • Reshaping: Adjust the dimensions of images as needed.
  • Normalization: Scale pixel values to a specific range, enhancing model performance.
  • Data Augmentation: Apply techniques to expand the dataset and increase model robustness.

This package aims to simplify and automate image preprocessing steps, optimizing the workflow in machine learning projects, the library link can be found here.

🔨 Installation

Use the package manager pip to install img-preprocess-pipeline

pip install img-preprocess-pipeline

Import

from image_preprocessing.preprocessing import tools, pipeline
from image_preprocessing.utils import io, display_images

⚙️ Functions

Load image

image_path = "your_image.png"
io.load_image(image_path,grayscale=False)

• Input:

  • image_path: directory of image
  • grayscale: load image with grayscale or RGB

• Output:

  • image (PIL Jpeg)

Description:

Load the image from the directory in scale RGB (with grayscale=False) and returns a tensor array.

Output example:

---

Save image

io.save_image(image,save_path) 

• Input:

  • image: choosen image file
  • save_path: directory to save the image

• Output:

  • None

Description:

Saves a image converting into a numpy array.

---

Reshape image

tools.reshape_image(image)

• Input:
  • image or dict: {title_of_image: image}
• Output:
  • dict: {'Reshaped': reshaped_image}

Description:

Reshapes a image and returns a dictionary with the title "Reshaped" and the resulting image.

Output example:

---

Scaling image

tools.scale_image(image)

• Input:
  • image or dict: {title_of_image:image}
• Output:
  • dict: {'Scaled':scaled_image}

Description: Scales a image (with range 0 to 1) and returns a dictionary with the title "Scaled" and the resulting image"""

Output example:

---

Random augumentation

tools.random_augumentation(image)

• Input:
  • image or dict: {title_of_image:image}
• Output:
  • dict: {'Augumented image':augumented_image}

Description: Create a new image modifying the saturation, brightness and rotation (in a random approach) of the original image and returns a dictionary with the title "Augumented image" and the resulting image.

Output example:

---

Preprocessing pipeline

pipeline.preprocess_pipeline(image) 

• Input:
  • image or dict: {title_of_image:image}
• Output:
  • dict: {'Preprocessed':preprocessed_image}

Description:

Create a new image using all the other functions in "tools" in combined into a pipeline. Returns a dictionary with the title "Preprocessed" and the resulting image.

Output example:

---

Display images

display_images.plot_images(image,image1,...,image_x) 

• Input:

  • dict of multiple images files. Ex:

    image1_with_label={'Original': image1}
    

• Output:

  • None

Description:

Displays the images and the respective labels of the images (Scaled,Preprocessed...) in a matrix of images with shape of (1,X) images.

Output example:

☕ Author and contact

Pablo Francisco Melo Bezerra

©️ License

MIT