GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecLocalDescriptor
Table of Contents ▼
Jump to a specific part of the page:
Description
LocalDescriptor holds interest point data and an optional local descriptor vector.
Attributes List
This module has the following attributes (case-insensitive ascending order):
Attributes
-
affineMatrix
(type:GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecMatrix2D
, default:nil
)
- Optional affine matrix. Supersedes scale and orientation if present. r' = affine_matrix.r + (x,y) defines an affine transform from the normalized image patch (in which the interest point is centered at the origin with scale 1) to the image. If the affine matrix is set, the following approximations are recommended: scale = sqrt(0.5 (xxxx + xyxy + yxyx + yyyy)); orientation = atan2(yx - xy, xx + yy); If not present, the affine matrix can be computed from scale and orientation as: xx = scale cos(orientation); xy = scale -sin(orientation); yx = scale sin(orientation); yy = scale * cos(orientation); -
data
(type:String.t
, default:nil
)
- -
dataFactor
(type:number()
, default:nil
)
- data_factor and data represent the local descriptor vector in a compressed format, using only 8 bit per value. Each byte of the data string yields one component of the local descriptor by bit-casting it to an int8 and multiplying it by data_factor. Protocol buffers do not support int8 directly. -
featureVector
(type:GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecFeatureVector
, default:nil
)
- Unquantized feature vector (float). -
opaqueData
(type:String.t
, default:nil
)
- Opaque descriptor data. May be used to pass through descriptor data from descriptor sources to processing modules, that is not already covered by data/data_factor and/or cannot be expressed as a vector of numbers. It is the responsibility of processing modules to verify that the data is in a compatible format. -
orientation
(type:number()
, default:nil
)
- Orientation is optional, as some interest point detectors don't compute it. The range of orientation is [-pi,pi). -
scale
(type:number()
, default:nil
)
- Each interest point must have a characteristic scale > 0. -
strength
(type:number()
, default:nil
)
- The strength or weight, indicating the relative significance of this point. -
x
(type:number()
, default:nil
)
- The position in the image with sub-pixel accuracy. The center of the upper left pixel has coordinates (0.0, 0.0). Thus the range for x and y is (-0.5, width - 0.5) x (-0.5, height - 0.5). -
y
(type:number()
, default:nil
)
-
Type
@type t() :: %GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecLocalDescriptor{
affineMatrix: GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecMatrix2D.t() | nil,
data: String.t() | nil,
dataFactor: number() | nil,
featureVector: GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecFeatureVector.t() | nil,
opaqueData: String.t() | nil,
orientation: number() | nil,
scale: number() | nil,
strength: number() | nil,
x: number() | nil,
y: number() | nil
}
affineMatrix: GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecMatrix2D.t() | nil,
data: String.t() | nil,
dataFactor: number() | nil,
featureVector: GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecFeatureVector.t() | nil,
opaqueData: String.t() | nil,
orientation: number() | nil,
scale: number() | nil,
strength: number() | nil,
x: number() | nil,
y: number() | nil
}
Function
@spec decode(struct(), keyword()) :: struct()Data sourced from HexDocs : GoogleApi.ContentWarehouse.V1.Model.PhotosVisionObjectrecLocalDescriptor