BaseObjectModelOutputConverter

class savant.base.converter.BaseObjectModelOutputConverter(**kwargs)

Base object model output converter.

abstract __call__(*output_layers, model, roi, metadata=None)

Converts raw model output tensors to a numpy array that represents a list of detected bboxes in the format (class_id, confidence, xc, yc, width, height, [angle]) in absolute coordinates computed with ROI information.

Parameters:

output_layers (ndarray | cupy.ndarray) – Model output layer tensors
model (ObjectModel) – Object model, required parameters: input tensor shape, maintain_aspect_ratio flag
roi (Tuple[float, float, float, float]) – [top, left, width, height] of the rectangle on which the model infers
metadata (NvDsFrameMeta | None) – Frame metadata. Optional for backward compatibility.

Returns:

BBox tensor (class_id, confidence, xc, yc, width, height, [angle]) offset by roi upper left and scaled by roi width and height

Return type:

ndarray | None

tensor_format: TensorFormat = 0: Set to CuPy to get the output_layers tensors in the converter call on the GPU as a cupy.ndarray. Or set to NumPy to get tensors on the host as a numpy.ndarray.