viam::sdk::MLModelService Class Reference

Represents a machine trained learning model instance. More...

#include "services/mlmodel/mlmodel.hpp"

Inheritance diagram for viam::sdk::MLModelService:

Classes
struct	metadata
struct	tensor_info

Public Types
template<typename T >
using	tensor_view = typename make_tensor_view_<T>::type
using	signed_integral_base_types
using	unsigned_integral_base_types
using	integral_base_types
using	fp_base_types = boost::mpl::list<float, double>
using	base_types = boost::mpl::joint_view<integral_base_types, fp_base_types>
using	tensor_view_types
using	tensor_views = boost::make_variant_over<tensor_view_types>::type
using	named_tensor_views = std::unordered_map<std::string, tensor_views>

Public Member Functions
API	api () const override
	Returns the API associated with a particular resource.
std::shared_ptr< named_tensor_views >	infer (const named_tensor_views &inputs)
	Runs the model against the input tensors and returns inference results as tensors.
virtual std::shared_ptr< named_tensor_views >	infer (const named_tensor_views &inputs, const ProtoStruct &extra)=0
	Runs the model against the input tensors and returns inference results as tensors.
struct metadata	metadata ()
	Returns metadata describing the inputs and outputs of the model.
virtual struct metadata	metadata (const ProtoStruct &extra)=0
	Returns metadata describing the inputs and outputs of the model.
Public Member Functions inherited from viam::sdk::Service
Name	get_resource_name () const override
	Returns the Name for a particular resource.
Public Member Functions inherited from viam::sdk::Resource
	Resource (std::string name)
virtual std::string	name () const
	Return the resource's name.

Static Public Member Functions
template<typename T >
static tensor_view< T >	make_tensor_view (const T *data, std::size_t size, typename tensor_view< T >::shape_type shape)

Protected Member Functions
	MLModelService (std::string name)
Protected Member Functions inherited from viam::sdk::Service
	Service (std::string name)
Protected Member Functions inherited from viam::sdk::Resource
Name	get_resource_name (const std::string &type) const

Detailed Description

Represents a machine trained learning model instance.

This class acts as an abstract base class to be used by any driver implementing an MLModel. It is also used as the base class for the gRPC client for communicating with a remote MLModel instance.

Member Typedef Documentation

integral_base_types

using viam::sdk::MLModelService::integral_base_types

Initial value:

 
        boost::mpl::joint_view<signed_integral_base_types, unsigned_integral_base_types>

signed_integral_base_types

using viam::sdk::MLModelService::signed_integral_base_types

Initial value:

 
        boost::mpl::list<std::int8_t, std::int16_t, std::int32_t, std::int64_t>

tensor_view_types

using viam::sdk::MLModelService::tensor_view_types

Initial value:

 
        boost::mpl::transform_view<base_types, make_tensor_view_<boost::mpl::placeholders::_1>>

unsigned_integral_base_types

using viam::sdk::MLModelService::unsigned_integral_base_types

Initial value:

 
        boost::mpl::transform_view<signed_integral_base_types,
                                   std::make_unsigned<boost::mpl::placeholders::_1>>

Member Function Documentation

api()

API viam::sdk::MLModelService::api ( ) const

overridevirtual

Returns the API associated with a particular resource.

Implements viam::sdk::Resource.

infer() [1/2]

std::shared_ptr< named_tensor_views > viam::sdk::MLModelService::infer ( const named_tensor_views & inputs )

inline

Runs the model against the input tensors and returns inference results as tensors.

Parameters

`inputs`

The input tensors on which to run inference.

Returns

The results of the inference as a shared pointer to named_tensor_views. The data viewed by the views is only valid for the lifetime of the returned shared_ptr.

infer() [2/2]

virtual std::shared_ptr< named_tensor_views > viam::sdk::MLModelService::infer ( const named_tensor_views & inputs, const ProtoStruct & extra )

pure virtual

Runs the model against the input tensors and returns inference results as tensors.

Parameters

`inputs`	The input tensors on which to run inference.
`extra`	Any additional arguments to the method.

Returns

The results of the inference as a shared pointer to named_tensor_views. The data viewed by the views is only valid for the lifetime of the returned shared_ptr.

Implemented in viam::sdk::impl::MLModelServiceClient, and viam::sdk::impl::MLModelServiceClient.

metadata()

virtual struct metadata viam::sdk::MLModelService::metadata ( const ProtoStruct & extra )

pure virtual

Returns metadata describing the inputs and outputs of the model.

Parameters

`extra`

Any additional arguments to the method.

Implemented in viam::sdk::impl::MLModelServiceClient, and viam::sdk::impl::MLModelServiceClient.

---

The documentation for this class was generated from the following file:

• src/viam/sdk/services/mlmodel.hpp