ConvInteger#

Domain: ai.onnx
Since version: 10

The integer convolution operator consumes an input tensor, its zero-point, a filter, and its zero-point, and computes the output. The production MUST never overflow. The accumulation may overflow if and only if in 32 bits.

Inputs

x (T1): Input data tensor from previous layer; has size (N x C x H x W), where N is the batch size, C is the number of channels, and H and W are the height and width. Note that this is for the 2D image. Otherwise the size is (N x C x D1 x D2 … x Dn). Optionally, if dimension denotation is in effect, the operation expects input data tensor to arrive with the dimension denotation of [DATA_BATCH, DATA_CHANNEL, DATA_FEATURE, DATA_FEATURE …].
w (T2): The weight tensor that will be used in the convolutions; has size (M x C/group x kH x kW), where C is the number of channels, and kH and kW are the height and width of the kernel, and M is the number of feature maps. For more than 2 dimensions, the kernel shape will be (M x C/group x k1 x k2 x … x kn), where (k1 x k2 x … kn) is the dimension of the kernel. Optionally, if dimension denotation is in effect, the operation expects the weight tensor to arrive with the dimension denotation of [FILTER_OUT_CHANNEL, FILTER_IN_CHANNEL, FILTER_SPATIAL, FILTER_SPATIAL …]. X.shape[1] == (W.shape[1] * group) == C (assuming zero based indices for the shape array). Or in other words FILTER_IN_CHANNEL should be equal to DATA_CHANNEL.
x_zero_point (T1): Zero point tensor for input ‘x’. It’s optional and default value is 0. It’s a scalar, which means a per-tensor/layer quantization.
w_zero_point (T2): Zero point tensor for input ‘w’. It’s optional and default value is 0. It could be a scalar or a 1-D tensor, which means a per-tensor/layer or per output channel quantization. If it’s a 1-D tensor, its number of elements should be equal to the number of output channels (M)

Outputs

y (T3): Output data tensor that contains the result of the convolution. The output dimensions are functions of the kernel size, stride size, and pad lengths.

Type Constraints

T1: Constrain input x and its zero point data type to 8-bit integer tensor. Allowed types: tensor(int8), tensor(uint8).
T2: Constrain input w and its zero point data type to 8-bit integer tensor. Allowed types: tensor(int8), tensor(uint8).
T3: Constrain output y data type to 32-bit integer tensor. Allowed types: tensor(int32).

Examples#

test_cc_basic_convinteger

Node:
  ConvInteger(X, W, x_zero_point) -> (Y)
  Attributes:
    kernel_shape = [2, 2]

Inputs:
  X: shape=(1, 1, 3, 3), dtype=uint8
    [[[[ 2,  3,  4],
       [ 5,  6,  7],
       [ 8,  9, 10]]]]
  W: shape=(1, 1, 2, 2), dtype=uint8
    [[[[1, 1],
       [1, 1]]]]
  x_zero_point: shape=(), dtype=uint8
    1

Outputs:
  Y: shape=(1, 1, 2, 2), dtype=int32
    [[[[12, 16],
       [24, 28]]]]

test_cc_convinteger_with_padding

Node:
  ConvInteger(X, W, x_zero_point) -> (Y)
  Attributes:
    kernel_shape = [2, 2]
    pads = [1, 1, 1, 1]

Inputs:
  X: shape=(1, 1, 3, 3), dtype=uint8
    [[[[ 2,  3,  4],
       [ 5,  6,  7],
       [ 8,  9, 10]]]]
  W: shape=(1, 1, 2, 2), dtype=uint8
    [[[[1, 1],
       [1, 1]]]]
  x_zero_point: shape=(), dtype=uint8
    1

Outputs:
  Y: shape=(1, 1, 4, 4), dtype=int32
    [[[[ 1,  3,  5,  3],
       [ 5, 12, 16,  9],
       [11, 24, 28, 15],
       [ 7, 15, 17,  9]]]]

test_cc_convinteger_without_padding

Node:
  ConvInteger(X, W, x_zero_point) -> (Y)

Inputs:
  X: shape=(1, 1, 3, 3), dtype=uint8
    [[[[ 2,  3,  4],
       [ 5,  6,  7],
       [ 8,  9, 10]]]]
  W: shape=(1, 1, 2, 2), dtype=uint8
    [[[[1, 1],
       [1, 1]]]]
  x_zero_point: shape=(), dtype=uint8
    1

Outputs:
  Y: shape=(1, 1, 2, 2), dtype=int32
    [[[[12, 16],
       [24, 28]]]]