Images

Image Objects

Images

Images(client, batch_size=100, response=None, **kwargs)

Bases: Entities

The python wrapper of images in ApertureDB.

This class serves 2 purposes:

This class is a layer on top of the native query.

It facilitate interactions with images in the database in a pythonic way. Abstracts the complexity of the query language and the communication with the database.

It includes utility methods to visualize image and annotations

Inter convert the representation into NumPy matrices and find similar images, related bounding boxes, etc.

Parameters:

Name	Type	Description	Default
client		The database connector, perhaps as returned by CommonLibrary.create_connector	required

total_results

total_results() -> int

Returns the total number of images that matched the query

Returns:

Name	Type	Description
int	int	Count of images that match the query.

get_image_by_index

get_image_by_index(index: int)

Get a single image by its index in the array of retrieved ids

Parameters:

Name	Type	Description	Default
index	int	Position in the image ids retrieved.	required

get_np_image_by_index

get_np_image_by_index(index: int)

Retrieves the NumPy representation of image from database

Parameters:

Name	Type	Description	Default
index	int	Position in the image ids retrieved.	required

get_bboxes_by_index

get_bboxes_by_index(index: int)

Get related bounding box for the image

Parameters:

Name	Type	Description	Default
index	int	Position in the image ids retrieved.	required

Returns:

Name	Type	Description
_type_		description

search

search(constraints=None, operations=None, format=None, limit=None, sort=None)

Sets a new query to retrieve images parameters as described below

Parameters:

Name	Type	Description	Default
constraints	dict	Constraints. Defaults to None.	None
operations	dict	Operations. Defaults to None.	None
format	str	Format of the returned images. Without this, the images are returned as they are stored. Defaults to None.	None
limit	int	number of values to restrict results to. Defaults to None.	None
sort	dict	Sort. Defaults to None.	None

search_by_property

search_by_property(prop_key: str, prop_values: list)

Constructs a constraints block and does a new search

Parameters:

Name	Type	Description	Default
prop_key	str	Key on which to search	required
prop_values	list	The values that must match for the key.	required

display

display(show_bboxes: bool = False, bbox_constraints: Constraints = None, show_polygons: bool = False, limit: Union[int, object] = None, polygon_constraints: Constraints = None, polygon_tag_key: str = '_label', polygon_tag_format: str = '{}') -> None

Display images with annotations

Parameters:

Name	Type	Description	Default
show_bboxes	bool	bool, optional Show bounding boxes, by default False	False
bbox_constraints	Constraints	Constraints, optional Constraints for bounding boxes, by default None	None
show_polygons	bool	bool, optional Show polygons, by default False	False
limit	Union[int, object]	Union[int, object], optional Number of images to display, by default None	None
polygon_constraints	Constraints	Constraints, optional Constraints for polygons, by default None	None
polygon_tag_key	str	str, optional Key for the polygon tag, by default "_label"	'_label'
polygon_tag_format	str	str, optional Format for the polygon tag, by default "{}"	'{}'

get_props_names

get_props_names() -> List[str]

Get the names of the properties that apply to images

Returns:

Name	Type	Description
properties	List[str]	The names of the properties of the images

get_properties

get_properties(prop_list: Iterable[str] = []) -> Dict[str, Any]

Get the properties of the images

Parameters:

Name	Type	Description	Default
prop_list	List[str]	The list of properties to retrieve. Defaults to [].	[]

Returns:

Name	Type	Description
property_values	Dict[str, Any]	The properties of the images

Frames

Frames(client, batch_size=100, response=None, **kwargs)

Bases: Images

The python wrapper of frame images in ApertureDB.

Frames in ApertureDB are quite similar to images and so are modeled in python as a subclass.

Parameters:

Name	Type	Description	Default
client		The database connector, perhaps as returned by CommonLibrary.create_connector	required

np_arr_img_to_bytes

np_arr_img_to_bytes(arr, format: str = 'JPEG') -> bytes

Convert a NumPy array to bytes

Parameters:

Name	Type	Description	Default
arr		The NumPy array to convert	required
format	str	The format of the image. Defaults to "JPEG".	'JPEG'

Returns:

Name	Type	Description
bytes	bytes	The image as bytes

image_to_bytes

image_to_bytes(image: Image, format: str = 'JPEG') -> bytes

Convert an image to bytes

Parameters:

Name	Type	Description	Default
image	Image	The image to convert	required
format	str	The format of the image. Defaults to "JPEG".	'JPEG'

Returns:

Name	Type	Description
bytes	bytes	The image as bytes

rotate

rotate(points, angle, c_x=0, c_y=0)

Rotate a set of points around a center

Parameters:

Name	Type	Description	Default
points		The points to rotate; iterable of (x, y) pairs	required
angle		The angle of counterclockwise rotation in degrees	required
c_x		The x coordinate of the center of rotation	0
c_y		The y coordinate of the center of rotation	0

Returns:

Name	Type	Description
points		The rotated points as a NumPy array of shape (n,2) and type int

resolve

resolve(points: array, image_meta, operations) -> np.array

Resolve the coordinates of a bounding box to the original image size

Given an array of (x,y) points with respect to the coordinates of an image, returns the corresponding coordinates with respect to a the resized or rotated image that resulted from a series of operations.

Parameters:

Name	Type	Description	Default
points	NumPy array	The coordinates of the bounding box; shape (...,2)	required
image_meta	dict	The image properties	required
operations	list	The operations applied to the image	required

Returns:

Name	Type	Description
resolved_points	NumPy array	The resolved coordinates, same shape as points