Skip to main content Link Search Menu Expand Document (external link)
Hidale.com Docs

@mlmodel attribute

🔌dp.kinect3 dp.oak
signature
mlmodel PATH_STRING
examples
@mlmodel c:\mymodels\squeezenet.onnx
@mlmodel darknet-abc123.onnx

Run machine learning inference with an ONNX model.

Input into model

Sensor color image is the single model input currently implemented. Pre-processing of this raw sensor data is controlled with @mlinput.

Output inference results

Model results must be dense tensors of numeric data. Multiple tensors are supported. Sparse tensors, sequences, and maps are not currently supported.

These results are output on the 5th outlet as a Max dictionary. route and routepass are helpful to send it to other dict objects for post-processing.

This dictionary contains a named entry for each result tensor output by the model. Their names are defined by the model itself.

{
    "$schema": "dp-nd-tensor",
    "shape": [ 1, 1000, 1, 1 ],
    "type": "float32",
    "data": [ 1.49474326e-05, 2.23620050e-03, 1.65242789e-04,
        9.92372283e-04, 7.42915319e-04, 5.80585969e-04,
        ...
    ]
}
  • $schema is “dp-nd-tensor”
  • shape is the tensor shape with an entry for every rank and its value as the extent
  • type is the tensor element type listed below
  • data contains the flattened element data of the tensor

Element types

ONNX data is converted from its native type into Max Dictionary compatible types like long, float32, char, etc.

Onnx Native Element Type C++ Type
ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT float32
ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT8 uint8
ONNX_TENSOR_ELEMENT_DATA_TYPE_INT8 int8
ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT16 uint16
ONNX_TENSOR_ELEMENT_DATA_TYPE_INT16 int16
ONNX_TENSOR_ELEMENT_DATA_TYPE_INT32 int32
ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64  
ONNX_TENSOR_ELEMENT_DATA_TYPE_STRING  
ONNX_TENSOR_ELEMENT_DATA_TYPE_BOOL  
ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT16 float16
ONNX_TENSOR_ELEMENT_DATA_TYPE_DOUBLE float64
ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT32  
ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT64  
ONNX_TENSOR_ELEMENT_DATA_TYPE_COMPLEX64  
ONNX_TENSOR_ELEMENT_DATA_TYPE_COMPLEX128  
ONNX_TENSOR_ELEMENT_DATA_TYPE_BFLOAT16