Resnet classification

ResNetClassification

Bases: ClassificationBaseOnnxRoboflowInferenceModel

VitClassification handles classification inference for Vision Transformer (ViT) models using ONNX.

Inherits

ClassificationBaseOnnxRoboflowInferenceModel: Base class for ONNX Roboflow Inference. ClassificationMixin: Mixin class providing classification-specific methods.

Attributes:

Name	Type	Description
multiclass	bool	A flag that specifies if the model should handle multiclass classification.

Source code in inference/models/resnet/resnet_classification.py

class ResNetClassification(ClassificationBaseOnnxRoboflowInferenceModel):
    """VitClassification handles classification inference
    for Vision Transformer (ViT) models using ONNX.

    Inherits:
        ClassificationBaseOnnxRoboflowInferenceModel: Base class for ONNX Roboflow Inference.
        ClassificationMixin: Mixin class providing classification-specific methods.

    Attributes:
        multiclass (bool): A flag that specifies if the model should handle multiclass classification.
    """

    preprocess_means = [0.485, 0.456, 0.406]
    preprocess_stds = [0.229, 0.224, 0.225]

    def __init__(self, *args, **kwargs):
        """Initializes the VitClassification instance.

        Args:
            *args: Variable length argument list.
            **kwargs: Arbitrary keyword arguments.
        """
        super().__init__(*args, **kwargs)
        self.multiclass = self.environment.get("MULTICLASS", False)

    @property
    def weights_file(self) -> str:
        """Determines the weights file to be used based on the availability of AWS keys.

        If AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY are set, it returns the path to 'weights.onnx'.
        Otherwise, it returns the path to 'best.onnx'.

        Returns:
            str: Path to the weights file.
        """
        if AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY and LAMBDA:
            return "weights.onnx"
        else:
            return "best.onnx"

weights_file property

Determines the weights file to be used based on the availability of AWS keys.

If AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY are set, it returns the path to 'weights.onnx'. Otherwise, it returns the path to 'best.onnx'.

Returns:

Name	Type	Description
str	str	Path to the weights file.

__init__(*args, **kwargs)

Initializes the VitClassification instance.

Parameters:

Name	Type	Description	Default
*args		Variable length argument list.	()
**kwargs		Arbitrary keyword arguments.	{}

Source code in inference/models/resnet/resnet_classification.py

def __init__(self, *args, **kwargs):
    """Initializes the VitClassification instance.

    Args:
        *args: Variable length argument list.
        **kwargs: Arbitrary keyword arguments.
    """
    super().__init__(*args, **kwargs)
    self.multiclass = self.environment.get("MULTICLASS", False)