Struct anira::v2::ProcessingSpec#

struct ProcessingSpec#

The 2.x processing specification: per input the channels and the samples per inference (0 = a non-streamable tensor), per output likewise and the model’s internal latency; the element counts are computed. An empty vector takes the 2.x default (1 channel, the per-channel element count, latency 0).

Public Functions

ProcessingSpec() = default#
inline ProcessingSpec(std::vector<size_t> preprocess_input_channels, std::vector<size_t> preprocess_output_channels, std::vector<size_t> preprocess_input_size, std::vector<size_t> postprocess_output_size, std::vector<size_t> internal_model_latency)#
inline ProcessingSpec(std::vector<size_t> preprocess_input_channels, std::vector<size_t> preprocess_output_channels)#
inline ProcessingSpec(std::vector<size_t> preprocess_input_channels, std::vector<size_t> preprocess_output_channels, std::vector<size_t> preprocess_input_size, std::vector<size_t> postprocess_output_size)#
inline bool operator==(const ProcessingSpec &other) const noexcept#

The five given vectors; the computed sizes are not compared (2.x).

inline bool operator!=(const ProcessingSpec &other) const noexcept#

Public Members

std::vector<size_t> m_preprocess_input_channels#
std::vector<size_t> m_postprocess_output_channels#
std::vector<size_t> m_preprocess_input_size#
std::vector<size_t> m_postprocess_output_size#
std::vector<size_t> m_internal_model_latency#
std::vector<size_t> m_tensor_input_size#

computed: the elements of each input

std::vector<size_t> m_tensor_output_size#

computed: the elements of each output