createFrom static method

TensorBuffer createFrom(TensorBuffer buffer, TfLiteType dataType)

Creates a TensorBuffer deep-copying data from another, with specified TfLiteType.

Throws ArgumentError.notNull if buffer is null.

Implementation

static TensorBuffer createFrom(TensorBuffer buffer, TfLiteType dataType) {
  SupportPreconditions.checkNotNull(buffer,
      message: "Cannot create a buffer from null");
  TensorBuffer result;
  if (buffer.isDynamic) {
    result = createDynamic(dataType);
  } else {
    result = createFixedSize(buffer.shape, dataType);
  }
  // The only scenario we need float array is FLOAT32->FLOAT32, or we can always use INT as
  // intermediate container.
  // The assumption is not true when we support other data types.
  if (buffer.getDataType() == TfLiteType.float32 &&
      dataType == TfLiteType.float32) {
    List<double> data = buffer.getDoubleList();
    result.loadList(data, shape: buffer.shape);
  } else {
    List<int> data = buffer.getIntList();
    result.loadList(data, shape: buffer.shape);
  }
  return result;
}