ultralytics_yolov8 module

This module implements the UltralyticsYOLOv8 wrapper class

The UltralyticsYOLOv8 class, which inherits from the ObjectDetectionModel interface, provides a wrapper for the YOLOv8 object detection model from the Ultralytics library. It enables you to use the YOLOv8 model within your program through the common interface defined in object_detection_model.py.

Key functionalities include:

• Provides a common interface for detections (via the __call__ method).

• Loading the YOLOv8 model from a weights file path.

• Performing object detection on an image using the YOLOv8 model.

• Converting the YOLOv8 model’s output to a list of Detection objects.

These Detection objects encapsulate details about detected objects, including bounding boxes, confidence scores, and potentially keypoints (if available in the model’s output).

This approach simplifies the integration and usage of YOLOv8 within this program, promoting code modularity and reusability.

class object_detection_models.ultralytics_yolov8.UltralyticsYOLOv8(model)

Bases: ObjectDetectionModel

Provides a wrapper for the YOLOv8 object detection model from the Ultralytics library.

This class inherits from the ObjectDetectionModel interface, enabling us to use the YOLOv8 model within our program through a consistent interface.

model

The underlying Ultralytics YOLOv8 model object.

static from_weights_path(weights_path: Path) → UltralyticsYOLOv8

Creates an UltralyticsYOLOv8 object from a path to the weights file.

Parameters:

weights_path (Path) – A path leading to the model’s weights.pt file.

Returns (UltralyticsYOLOv8):

An UltralyticsYOLOv8 object.

static from_model(model) → UltralyticsYOLOv8

Creates an UltralyticsYOLOv8 object from the Ultralytics model object.

Parameters:

model – The Ultralytics class for the YOLOv8 model.

yolov8_results_to_detections(results) → List[Detection]

Converts ultralytics’ YOLOv8 model object’s results to a list of Detection objects.

Parameters:

results – List containing the output from a YOLOv8 model prediction. Refer to the YOLOv8 documentation for details on the output format.

Returns:

A list of Detection objects. Each Detection object contains information about a detected object including its bounding box (category, coordinates), and confidence score. Additionally, if keypoints are present in the results, they are added to the Detection objects.

Raises:

Exception – If an error occurs during processing of the results (e.g., keypoints are not found).