NvInferObjectModelOutput

class savant.deepstream.nvinfer.model.NvInferObjectModelOutput(layer_names=<factory>, converter=None, objects=None, num_detected_classes=None)

NvInferObjectModel output configuration template for detector with aligned bboxes.

Example

model:
    # model configuration
    output:
        num_detected_classes: 4
        layer_names: [output]
        objects:
            # output objects configuration

converter: Optional[PyFunc] = None

Model output converter. Converter is used to transform raw tensor output into Savant data format.

Converter implementation should be written as a subclass of BaseObjectModelOutputConverter or BaseAttributeModelOutputConverter or BaseComplexModelOutputConverter depending on the model type.

layer_names: List[str]: Specify model output layer names.

objects: Optional[List[NvInferObjectModelOutputObject]] = None: Configuration for each output object class of an object model.

num_detected_classes: Optional[int] = None: Number of classes detected by the model. Required for regular detector.