YoloDataLabeler V1.0.2

🎉 Release Notes for V1.0.2 🎉

🏗 Architectural Overhaul

• MVP Transition: We've elevated the architecture of our application by transitioning to the robust Model-View-Presenter (MVP) structure. This shift ensures a more modular, maintainable, and testable codebase, setting a solid foundation for future features and improvements.

📊 Enhanced Data Labeling Capabilities

• Stratified Data Splitting: With an aim to enhance accuracy in data labeling, we've incorporated the stratify feature from scikit-learn's train_test_split. This ensures that the resulting splits have a consistent representation of target labels, leading to better and more reliable labeling outcomes.

🐞 Quality & Maintenance

• Bug Exterminations: We've identified and addressed several bugs from previous releases, ensuring a smoother and more reliable user experience.

🛠 Under The Hood

• Built on PyQt5: Our commitment to delivering a seamless UI experience continues with the integration of PyQt5, offering users an intuitive and efficient data labeling journey.

YoloDataLabeler V1.0.0

YOLO Data Labeler v1.0.0 Release Notes

Introduction

We are thrilled to introduce the YOLO Data Labeler! Developed by Melih Taşkın, this tool is designed with an emphasis on ease-of-use for labeling datasets, specifically optimized for the YOLO model. Dive into a streamlined experience and prepare your YOLO dataset without any hassle.

Key Features

• Easy Image Selection: Initiate your labeling process effortlessly by simply pointing to your images folder.
• JSON Label Import: Bypass the tedious manual labeling with a straightforward JSON structure:

{
"0":"cat",
"1":"dog",
}

• Image Resizing: Adjust image dimensions with ease. Input your desired width and height, and the app will do the rest.
• Dataset Creation: Break your data down into training, validation, and test sets seamlessly. Define your destination folder and set percentages, then let the tool do the heavy lifting!

Data Privacy & Security 🛡️

We recognize the paramount importance of data security. YOLO Data Labeler champions data privacy:

• Local Operations: All labeling tasks are exclusively local, ensuring no data uploads.
• Full Data Autonomy: Your data remains untouched and in your hands, ensuring utmost confidentiality.
• Transparent Processing: We guarantee no covert backend processes. Your data is and will always be solely yours.

Getting Started

Windows Users:

• Start either with our pre-built executable from the dist folder or use Python after navigating to the repository root and executing python main.py.

Unix/Mac Users:

• We're aware of a minor issue affecting the MacOS and Unix releases. Our team is actively working on it and we aim to release a fix very soon.
• We appreciate your patience and understanding. Stay tuned for updates!

Engage with Us!

Your insights drive improvement. We eagerly await your feedback, bug reports, and contributions. For significant updates or enhancements, please initiate an issue for a focused discussion.

License & Acknowledgments

• Licensed under MIT License. Check out the LICENSE file for detailed information.
• Gratitude to the open-source aficionados and the pioneers behind the YOLO model.
• Special mention to Ahmet Koyuncu for his unwavering support and indispensable contributions.