Yolov5 instance segmentation

YOLOv5InstanceSegmentation

Bases: InstanceSegmentationBaseOnnxRoboflowInferenceModel

YOLOv5 Instance Segmentation ONNX Inference Model.

This class is responsible for performing instance segmentation using the YOLOv5 model with ONNX runtime.

Attributes:

Name	Type	Description
weights_file	str	Path to the ONNX weights file.

Source code in inference/models/yolov5/yolov5_instance_segmentation.py

class YOLOv5InstanceSegmentation(InstanceSegmentationBaseOnnxRoboflowInferenceModel):
    """YOLOv5 Instance Segmentation ONNX Inference Model.

    This class is responsible for performing instance segmentation using the YOLOv5 model
    with ONNX runtime.

    Attributes:
        weights_file (str): Path to the ONNX weights file.
    """

    @property
    def weights_file(self) -> str:
        """Gets the weights file for the YOLOv5 model.

        Returns:
            str: Path to the ONNX weights file.
        """
        return "yolov5s_weights.onnx"

    def predict(self, img_in: np.ndarray, **kwargs) -> Tuple[np.ndarray, np.ndarray]:
        """Performs inference on the given image using the ONNX session.

        Args:
            img_in (np.ndarray): Input image as a NumPy array.

        Returns:
            Tuple[np.ndarray, np.ndarray]: Tuple containing two NumPy arrays representing the predictions.
        """
        with self._session_lock:
            predictions = run_session_via_iobinding(
                self.onnx_session, self.input_name, img_in
            )
        return predictions[0], predictions[1]

weights_file property

Gets the weights file for the YOLOv5 model.

Returns:

Name	Type	Description
str	str	Path to the ONNX weights file.

predict(img_in, **kwargs)

Performs inference on the given image using the ONNX session.

Parameters:

Name	Type	Description	Default
img_in	ndarray	Input image as a NumPy array.	required

Returns:

Type	Description
Tuple[ndarray, ndarray]	Tuple[np.ndarray, np.ndarray]: Tuple containing two NumPy arrays representing the predictions.

Source code in inference/models/yolov5/yolov5_instance_segmentation.py

def predict(self, img_in: np.ndarray, **kwargs) -> Tuple[np.ndarray, np.ndarray]:
    """Performs inference on the given image using the ONNX session.

    Args:
        img_in (np.ndarray): Input image as a NumPy array.

    Returns:
        Tuple[np.ndarray, np.ndarray]: Tuple containing two NumPy arrays representing the predictions.
    """
    with self._session_lock:
        predictions = run_session_via_iobinding(
            self.onnx_session, self.input_name, img_in
        )
    return predictions[0], predictions[1]