NvCvTensorOpsExtension¶
Generic CVCORE tensor_ops interfaces.
UUID: 6eae64ff-97a9-4d9b-b324-f85e6a98a75a
Version: 1.0.0
Author: NVIDIA
License: NVIDIA
Components¶
nvidia::cvcore::tensor_ops::Resize¶
Codelet for image resizing in tensor_ops.
Component ID: a58141ac-7eca-4ea5-9b54-5446fe379a11
Base Type: nvidia::gxf::Codelet
Parameters¶
input_type
Describes the input image type expected at the receiver.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
output_width
Width of the output tensor.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_UINT64
output_height
Height of the output tensor.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_UINT64
interp_type
Interpolation type for resizing (nearest/linear/cubic_bspline/cubic_catmullrom).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
keep_aspect_ratio
Whether to keep aspect ratio while resizing.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_BOOL
receiver
Named receiver queue with the input to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
transmitter
Named transmitter to send the output to.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Transmitter
pool
Named memory pool to allocate blocks from. (must be GPU pool).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Allocator
output_name
Name of the output tensor.
Flags: GXF_PARAMETER_FLAGS_OPTIONAL
Type: GXF_PARAMETER_TYPE_STRING
nvidia::cvcore::tensor_ops::CropAndResize¶
Codelet for crop and resizing operation in tensor_ops.
Component ID: 4a7ff422-de38-41bc-9e74-3ac10d9294b6
Base Type: nvidia::gxf::Codelet
Parameters¶
input_type
Describes the input image type expected at the receiver.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
output_width
Width of the output tensor.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_UINT64
output_height
Height of the output tensor.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_UINT64
interp_type
Interpolation type for resizing (nearest/linear/cubic_bspline/cubic_catmullrom).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
keep_aspect_ratio
Whether to keep aspect ratio while resizing.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_BOOL
receiver
Named receiver queue with the input to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
receiver_bbox
Named receiver queue with the input bbox to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
transmitter
Named transmitter to send the output to.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Transmitter
pool
Named memory pool to allocate blocks from. (Must be GPU pool.)
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Allocator
output_name
Name of the output tensor.
Flags: GXF_PARAMETER_FLAGS_OPTIONAL
Type: GXF_PARAMETER_TYPE_STRING
nvidia::cvcore::tensor_ops::Normalize¶
Codelet for image normalization in tensor_ops.
Component ID: 7018f0b9-034c-462b-a9fb-af7ee012974f
Base Type: nvidia::gxf::Codelet
Parameters¶
input_type
Describes the input image type expected at the receiver.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
scales
Scaling factors used for normalization (must be the same size of channel count).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_CUSTOM
offsets
Offset factors used for normalization.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_CUSTOM
receiver
Named receiver queue with the input to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
transmitter
Named transmitter to send the output to.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Transmitter
pool
Named memory pool to allocate blocks from (must be GPU pool).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Allocator
output_name
Name of the output tensor
Flags: GXF_PARAMETER_FLAGS_OPTIONAL
Type: GXF_PARAMETER_TYPE_STRING
nvidia::cvcore::tensor_ops::InterleavedToPlanar¶
Codelet for convert interleaved image to planar image in tensor_ops.
Component ID: 269d4237-f3c3-479d-bcca-9ecc44c71a70
Base Type: nvidia::gxf::Codelet
Parameters¶
input_type
Describes the input image type expected at the receiver.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
output_width
Width of the output tensor.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_UINT64
output_height
Height of the output tensor.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_UINT64
interp_type
Interpolation type for resizing (nearest/linear/cubic_bspline/cubic_catmullrom).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
keep_aspect_ratio
Whether to keep aspect ratio while resizing.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_BOOL
receiver
Named receiver queue with the input to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
transmitter
Named transmitter to send the output to.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Transmitter
pool
Named memory pool to allocate blocks from (must be GPU pool).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Allocator
output_name
Name of the output tensor.
Flags: GXF_PARAMETER_FLAGS_OPTIONAL
Type: GXF_PARAMETER_TYPE_STRING
nvidia::cvcore::tensor_ops::ConvertColorFormat¶
Codelet for image color conversion in tensor_ops.
Component ID: fc4d7b4d-8fcc-4daa-a286-056e0fcafa78
Base Type: nvidia::gxf::Codelet
Parameters¶
input_type
Describes the input image type expected at the receiver.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
output_type
Describes the output image type.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
receiver
Named receiver queue with the input to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
transmitter
Named transmitter to send the output to.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Transmitter
pool
Named memory pool to allocate blocks from (must be GPU pool).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Allocator
output_name
Name of the output tensor.
Flags: GXF_PARAMETER_FLAGS_OPTIONAL
Type: GXF_PARAMETER_TYPE_STRING
nvidia::cvcore::tensor_ops::Reshape¶
Codelet for image reshape in tensor_ops.
Component ID: 26789b7d-5a8d-4e84-86b8-45ec5f4cd12a
Base Type: nvidia::gxf::Codelet
Parameters¶
input_type
Describes the input image type expected at the receiver.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_STRING
output_shape
Output tensor shape (must have the same size as input Tensor).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_CUSTOM
receiver
Named receiver queue with the input to the pipeline.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Receiver
transmitter
Named transmitter to send the output to.
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Transmitter
pool
Named memory pool to allocate blocks from (must be GPU pool).
Flags: GXF_PARAMETER_FLAGS_NONE
Type: GXF_PARAMETER_TYPE_HANDLE
Handle Type: gxf::Allocator
output_name
Name of the output tensor.
Flags: GXF_PARAMETER_FLAGS_OPTIONAL
Type: GXF_PARAMETER_TYPE_STRING