Datatype: The underlying storage for a Tensor. More...

#include <rerun/datatypes/tensor_buffer.hpp>

Public Member Functions
	TensorBuffer (const TensorBuffer &other)
	Copy constructor.

TensorBuffer &	operator= (const TensorBuffer &other) noexcept

	TensorBuffer (TensorBuffer &&other) noexcept

TensorBuffer &	operator= (TensorBuffer &&other) noexcept

	TensorBuffer (Collection< uint8_t > u8)
	Construct a `TensorBuffer` from a `Collection<uint8_t>`.

	TensorBuffer (Collection< uint16_t > u16)
	Construct a `TensorBuffer` from a `Collection<uint16_t>`.

	TensorBuffer (Collection< uint32_t > u32)
	Construct a `TensorBuffer` from a `Collection<uint32_t>`.

	TensorBuffer (Collection< uint64_t > u64)
	Construct a `TensorBuffer` from a `Collection<uint64_t>`.

	TensorBuffer (Collection< int8_t > i8)
	Construct a `TensorBuffer` from a `Collection<int8_t>`.

	TensorBuffer (Collection< int16_t > i16)
	Construct a `TensorBuffer` from a `Collection<int16_t>`.

	TensorBuffer (Collection< int32_t > i32)
	Construct a `TensorBuffer` from a `Collection<int32_t>`.

	TensorBuffer (Collection< int64_t > i64)
	Construct a `TensorBuffer` from a `Collection<int64_t>`.

	TensorBuffer (Collection< rerun::half > f16)
	Construct a `TensorBuffer` from a `Collection<half>`.

	TensorBuffer (Collection< float > f32)
	Construct a `TensorBuffer` from a `Collection<float>`.

	TensorBuffer (Collection< double > f64)
	Construct a `TensorBuffer` from a `Collection<double>`.

template<typename TContainer , typename value_type = traits::value_type_of_t<TContainer>>
	TensorBuffer (TContainer &&container)
	Construct a `TensorBuffer` from any container type that has a `value_type` member and for which a `rerun::ContainerAdapter` exists.

size_t	num_elems () const
	Number of elements in the buffer.

void	swap (TensorBuffer &other) noexcept

const rerun::Collection< uint8_t > *	get_u8 () const
	Return a pointer to u8 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< uint16_t > *	get_u16 () const
	Return a pointer to u16 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< uint32_t > *	get_u32 () const
	Return a pointer to u32 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< uint64_t > *	get_u64 () const
	Return a pointer to u64 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< int8_t > *	get_i8 () const
	Return a pointer to i8 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< int16_t > *	get_i16 () const
	Return a pointer to i16 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< int32_t > *	get_i32 () const
	Return a pointer to i32 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< int64_t > *	get_i64 () const
	Return a pointer to i64 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< rerun::half > *	get_f16 () const
	Return a pointer to f16 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< float > *	get_f32 () const
	Return a pointer to f32 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< double > *	get_f64 () const
	Return a pointer to f64 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< uint8_t > *	get_jpeg () const
	Return a pointer to jpeg if the union is in that state, otherwise `nullptr`.

const rerun::Collection< uint8_t > *	get_nv12 () const
	Return a pointer to nv12 if the union is in that state, otherwise `nullptr`.

const rerun::Collection< uint8_t > *	get_yuy2 () const
	Return a pointer to yuy2 if the union is in that state, otherwise `nullptr`.

Static Public Member Functions
static TensorBuffer	u8 (rerun::Collection< uint8_t > u8)
	8bit unsigned integer.

static TensorBuffer	u16 (rerun::Collection< uint16_t > u16)
	16bit unsigned integer.

static TensorBuffer	u32 (rerun::Collection< uint32_t > u32)
	32bit unsigned integer.

static TensorBuffer	u64 (rerun::Collection< uint64_t > u64)
	64bit unsigned integer.

static TensorBuffer	i8 (rerun::Collection< int8_t > i8)
	8bit signed integer.

static TensorBuffer	i16 (rerun::Collection< int16_t > i16)
	16bit signed integer.

static TensorBuffer	i32 (rerun::Collection< int32_t > i32)
	32bit signed integer.

static TensorBuffer	i64 (rerun::Collection< int64_t > i64)
	64bit signed integer.

static TensorBuffer	f16 (rerun::Collection< rerun::half > f16)
	16bit IEEE-754 floating point, also known as `half`.

static TensorBuffer	f32 (rerun::Collection< float > f32)
	32bit IEEE-754 floating point, also known as `float` or `single`.

static TensorBuffer	f64 (rerun::Collection< double > f64)
	64bit IEEE-754 floating point, also known as `double`.

static TensorBuffer	jpeg (rerun::Collection< uint8_t > jpeg)
	Raw bytes of a JPEG file.

static TensorBuffer	nv12 (rerun::Collection< uint8_t > nv12)
	NV12 is a YUV 4:2:0 chroma downsamples format with 8 bits per channel.

static TensorBuffer	yuy2 (rerun::Collection< uint8_t > yuy2)
	YUY2, also known as YUYV is a YUV 4:2:2 chroma downsampled format with 8 bits per channel.

Detailed Description

Datatype: The underlying storage for a Tensor.

Tensor elements are stored in a contiguous buffer of a single type.

Constructor & Destructor Documentation

◆ TensorBuffer()

template<typename TContainer , typename value_type = traits::value_type_of_t<TContainer>>

rerun::datatypes::TensorBuffer::TensorBuffer ( TContainer && container )

inline

Construct a TensorBuffer from any container type that has a value_type member and for which a rerun::ContainerAdapter exists.

This constructor assumes the type of tensor buffer you want to use is defined by TContainer::value_type and then forwards the argument as-is to the appropriate rerun::Container constructor.

See also: rerun::ContainerAdapter, rerun::Container

Member Function Documentation

◆ num_elems()

size_t rerun::datatypes::TensorBuffer::num_elems ( ) const

Number of elements in the buffer.

You may NOT call this for JPEG buffers.

◆ nv12()

static TensorBuffer rerun::datatypes::TensorBuffer::nv12 ( rerun::Collection< uint8_t > nv12 )

inlinestatic

NV12 is a YUV 4:2:0 chroma downsamples format with 8 bits per channel.

First comes entire image in Y, followed by interleaved lines ordered as U0, V0, U1, V1, etc.

◆ yuy2()

static TensorBuffer rerun::datatypes::TensorBuffer::yuy2 ( rerun::Collection< uint8_t > yuy2 )

inlinestatic

YUY2, also known as YUYV is a YUV 4:2:2 chroma downsampled format with 8 bits per channel.

The order of the channels is Y0, U0, Y1, V0.

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

rerun/datatypes/tensor_buffer.hpp

Public Member Functions