Bases: NumpyScalar
The local features found in one image.
xy: pixel centres, expressed at the stored image's size whatever
size the detector ran at.
scale: feature scales in pixels.
angle: orientations in degrees.
descriptors: 128-byte descriptors.
color: RGB sampled from the image at each centre.
width, height: the image size xy is expressed at.
Analysis-only: produced by a recipe, never captured, so it has no platform
mirror and no place in a schema.
Source code in capturegraph-recipes/capturegraph_recipes/types/reconstruction/keypoints.py
| class Keypoints(cg.NumpyScalar, name="edu.cornell.keypoints"):
"""The local features found in one image.
- ``xy``: pixel centres, expressed at the stored image's size whatever
size the detector ran at.
- ``scale``: feature scales in pixels.
- ``angle``: orientations in degrees.
- ``descriptors``: 128-byte descriptors.
- ``color``: RGB sampled from the image at each centre.
- ``width``, ``height``: the image size ``xy`` is expressed at.
Analysis-only: produced by a recipe, never captured, so it has no platform
mirror and no place in a schema.
"""
__slots__ = ()
analysis_only = True
layout = {
"xy": ("float32", (-1, 2)),
"scale": ("float32", (-1,)),
"angle": ("float32", (-1,)),
"descriptors": ("uint8", (-1, 128)),
"color": ("uint8", (-1, 3)),
"width": ("int32", ()),
"height": ("int32", ()),
}
|