holoscan::ops::BufferInfo | Holoscan SDK Documentation

Buffer information, can be initialized either with a tensor or a video buffer.

#include <holoscan/ops/buffer_info.hpp>

Methods

init

Initialize with tensor

Initialize with video buffer

gxf_result_t holoscan::ops::BufferInfo::init(
    const nvidia::gxf::Handle<nvidia::gxf::Tensor> &tensor,
    HolovizOp::ImageFormat input_image_format = HolovizOp::ImageFormat::AUTO_DETECT
)

Initialize with tensor.

Returns: error code

Parameters

input_image_format

HolovizOp::ImageFormatDefaults to HolovizOp::ImageFormat::AUTO_DETECT

If HolovizOp::ImageFormat::AUTO_DETECT then auto detect the image format, else use the provided image format

Static methods

get_supported_tensor_formats_str

static std::string holoscan::ops::BufferInfo::get_supported_tensor_formats_str(
    const std::vector<HolovizOp::ImageFormat> &device_image_formats
)

Returns: a string describing the supported tensor formats

get_supported_video_buffer_formats_str

static std::string holoscan::ops::BufferInfo::get_supported_video_buffer_formats_str(
    const std::vector<HolovizOp::ImageFormat> &device_image_formats
)

Returns: a string describing the supported video buffer formats

Member variables

Name	Type	Description
`rank`	`uint32_t`
`components`	`uint32_t`
`width`	`uint32_t`
`height`	`uint32_t`
`element_type`	`nvidia::gxf::PrimitiveType`
`image_format`	`HolovizOp::ImageFormat`
`component_swizzle`	`viz::ComponentSwizzle`
`name`	`std::string`
`buffer_ptr`	`const nvidia::byte *`	points to the memory owned by either a tensor or video buffer
`storage_type`	`nvidia::gxf::MemoryStorageType`
`bytes_size`	`uint64_t`
`stride`	`nvidia::gxf::Tensor::stride_array_t`
`color_planes`	`std::vector< nvidia::gxf::ColorPlane >`	plane information
`yuv_model_conversion`	`HolovizOp::YuvModelConversion`
`yuv_range`	`HolovizOp::YuvRange`