Skip to content

Latest commit

 

History

History
95 lines (65 loc) · 3.51 KB

readme.md

File metadata and controls

95 lines (65 loc) · 3.51 KB

Image Organizer using ResNet-50

This project organizes images into folders based on visual similarity using a pre-trained ResNet-50 model.

Table of Contents

Introduction

Do you ever find your screenshots folder cluttered and disorganized? This project aims to solve that problem by automatically organizing images into folders based on their visual similarity. Using a pre-trained ResNet-50 model, this script analyzes the content of your images and sorts them into appropriate folders, making it easier to find and manage your screenshots.

Installation

  1. Clone the repository:

    git clone https://github.com/arjunj05/ScreenshotSorter.git
cd ScreenshotSorter

  2. Set up a virtual environment:

    python3 -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`

  3. Install the dependencies:

    pip install -r requirements.txt

Usage

  1. Run the script:

    python ScreenshotOrganizer.py

  2. Enter the paths when prompted:

    • Path of folders to organize into.
    • Path of screenshots to be sorted.

Features

  • Uses a pre-trained ResNet-50 model to extract image embeddings.
  • Calculates cosine similarity to organize images based on visual similarity.
  • Automatically creates new folders for images that do not fit existing categories.

Technical Details

  • Model: ResNet-50 pre-trained on ImageNet.
  • Libraries: PyTorch, torchvision, PIL, numpy.
  • Image Processing: Resizing, normalizing, and tensor conversion.

What I Learned

  • Image Processing Techniques: Gained hands-on experience with image preprocessing techniques such as resizing, normalization, and tensor conversion, essential for preparing images for machine learning models.

  • Deep Learning Models: Improved understanding of how pre-trained deep learning models, specifically ResNet-50, can be used for feature extraction and how to leverage them for various tasks beyond classification.

  • Feature Extraction and Embeddings: Learned about generating and using embeddings for images, which involves extracting high-level features that capture the essence of an image's content for similarity comparison.

  • Cosine Similarity for Image Matching: Implemented and understood the concept of cosine similarity to measure the similarity between image embeddings, which is crucial for organizing images based on visual similarity.

  • Performance Optimization: Gained insights into optimizing performance by caching embeddings to avoid redundant computations and improve the efficiency of the image organization process.

Contributing

Contributions are welcome! Please follow these steps:

  1. Fork the repository.
  2. Create a new branch (git checkout -b feature-branch).
  3. Commit your changes (git commit -am 'Add new feature').
  4. Push to the branch (git push origin feature-branch).
  5. Create a new Pull Request.

Contact Information

For any questions or suggestions, please contact me at [email protected].

Acknowledgments