E2E Video and Image Preprocessing for DL: Domain Independent Pipeline

📖 Description

e2evideo is a versatile Python package designed for video and image pre-processing and analysis 🎥📸. It comprises domain-independent modules that can be customized to suit specific tasks in various fields of computer vision.

🛠️ Installation

To install e2evideo, clone the Git repository, navigate to the directory, and run:

pip install .

🚀 Features :

• 🎞️ Video Pre-processing: Supports various video formats, frame extraction, and background subtraction.

• 🖼️ Image Pre-processing: Converts images to greyscale, resizes images, and structures videos into compressed arrays.

• 🧠 Feature Extraction: Includes a pre-trained ResNet18 model and DINOv2 for image embedding extraction.

💻 Usage

Import the package and utilize its modules as required:

import e2evideo
# Your code here

📚 Documentation

For more detailed instructions and examples, refer to the Documentation.

🤝 Contributing

Contributions to E2Evideo are welcome! If you would like to contribute, please fork the repository and create a pull request.

📜 License

E2Evideo is available under the MIT License 📄.

📃 Citation

For academic use, please cite the package as follows:

@inproceedings{10.1007/978-3-031-53302-0_19,
    author = {Alawad, Faiga and Halvorsen, P{\aa}l and Riegler, Michael A.},
    booktitle = {MultiMedia Modeling},
    editor = {Rudinac, Stevan and Hanjalic, Alan and Liem, Cynthia and Worring, Marcel and J{\'o}nsson, Bj{\"o}rn Þ{\'o}r and Liu, Bei and Yamakata, Yoko},
    isbn = {978-3-031-53302-0},
    pages = {258--264},
    publisher = {Springer Nature Switzerland},
    title = {E2Evideo: End to End Video and Image Pre-processing and Analysis Tool},
    year = {2024}}