Odlabel

Revise label figure naming convention to enhance clarity in the output chart.

Release Notes: Improved Drawing of Bounding Boxes and Labels

We are excited to announce the release of an enhanced version of the `draw_boxes_on_image` function, which now includes intelligent scaling of bounding boxes and labels based on the size of the detected objects.

Key Features

1. **Adaptive Font Scaling**: The font size of the class labels is now automatically adjusted based on the size of the detected objects. Larger objects will have larger font sizes, ensuring that the labels remain readable and proportional to the bounding box size.

2. **Dynamic Text Thickness**: In addition to the font size, the thickness of the label text is also dynamically adjusted based on the object size. This ensures that the labels maintain good visibility and clarity regardless of the object's dimensions.

3. **Consistent Color Scheme**: The function now utilizes a `class_colors` dictionary to assign consistent colors to each class ID across multiple images. This enhances visual coherence and makes it easier to identify and track objects of the same class.

4. **Improved Background Rectangle**: The background rectangle behind the class labels has been enhanced to provide better readability. The rectangle size is intelligently adjusted based on the label text dimensions, ensuring that the text fits comfortably within the background.

We are excited to announce the initial release of ODLabel, a powerful tool for zero-shot object detection, labeling, and visualization. ODLabel provides an intuitive graphical user interface that enables users to efficiently label objects in images using the YOLO-World model.

Key features in this release:

- Support for selecting from various YOLO-World model options, including yolov8s-world, yolov8m-world, yolov8l-world, and yolov8x-world.
- Ability to choose an images folder for labeling and specify an output directory for the annotated data.
- Flexibility to define the object categories you want to detect.
- Integration of Slicing Adaptive Inference (SAHI) for improved detection of small objects.
- Option to select the device type (CPU or GPU) for inference.
- Customization of the train/validation split ratio.
- Adjustment of confidence level and non-maximum suppression (IoU) threshold.
- Comprehensive dashboard with figures and visualizations to explore input image data and detection results.

We believe ODLabel will be a valuable tool for researchers, developers, and anyone working on object detection and labeling tasks. This initial release lays the foundation for a powerful and user-friendly application, and we look forward to receiving feedback and continued improvement in future updates.

For detailed installation and usage instructions, please refer to the [project's README](https://github.com/Ziad-Algrafi/ODLabel/edit/main/README.md).

We hope you find ODLabel helpful in your work! If you have any questions or feedback, feel free to reach out to us at ZiadAlgrafigmail.com.