Converts the array of LocatedObject instances to a report.

Merges a grid of sub-images into a single image, including the annotations.
Opposite operation to adams.data.image.transformer.subimages.Grid.
Expects the image array to represent the cells in the grid in a row-wise fashion.
Only combines object annotations from reports, other field values will be discarded.
Does not remove overlapping objects, see adams.flow.transformer.DeleteOverlappingImageObjects.

See also:
adams.data.image.transformer.subimages.Grid
adams.flow.transformer.DeleteOverlappingImageObjects

Ancestor for classes that generate subimages from a single image.

Extracts the sub-images according to the region definitions.
Additional report values:
- Region: for the region
- Coordinates: for the 1-based coordinates (x,y,w,h)

Dummy generator, simply passes the original image through.

Loads COCO annotations from the JSON file, with one report per image.
Handles only segmentations with polygons (not RLE) and only one polygon per annotation.

Reads object locations from a spreadsheet into a report.
Top/left column is required.
Either right/bottom or width/height need to be supplied.
In addition, polygon coordinates (X and Y coordinates as comma-separated lists in two separate columns) can be read as well.
If the coordinates/dimensions represent normalized ones (ie 0-1), then specify the width/height of the image to relate them back to actual pixel-based sizes.

Reads object locations in Pascal VOC format.

See more:
https://github.com/WaikatoLink2020/objdet-predictions-exchange-format

Interprets polygon annotations generated by VGG Image Annotator.
For more information, see:
http://www.robots.ox.ac.uk/~vgg/software/via/

Writes point annotations from multiple reports in DeepLabCut's CSV format.
Expects the filename to be stored in the 'File' report field.

Interface for finders that locate objects in the report of an image.

Ancestor for object finders that use a base finder.

Returns all indices.

Outputs the indices of objects where the specified regular expression matches the value associated with the given meta-data key.

Inverts the indices of the base object finder.

Finds no indices.

Locates the objects that fall into the defined region.
Partial hits can be considered as well.

Applies the minimum/maximum constraints to width/height (according to the chosen type).

Compares the object types between objects from the two reports.
The first report is considered 'ground truth'.

Counts the objects in the report passing through that fall into the defined region.
Partial hits can be counted as well, using their percentage in overlap as count.

Outputs the requested type of information for either the incoming adams.flow.transformer.locateobjects.LocatedObject or the specified image object in the report.

Merges the object locations in the report of the container passing through with the one obtained from storage.
The 'overlap action' determines what to do if objects overlap.
With the 'check type' you can still trigger a 'skip' if the type values of the two overlapping objects differ.
For simply merging all objects, choose a minOverlapRatio of 0.0 and the OverlapAction of KEEP.

Scales the objects in the report using the provided scale factors.
Processes the following suffixes of all the report fields that match the provided prefix:
- .x
- .y
- .width
- .height

Input/output:
- accepts:
   adams.data.report.Report
   adams.data.report.ReportHandler
- generates:
   adams.data.report.Report
   adams.data.report.ReportHandler

Overlays object annotations from the report.

Allows annotating images and/or modifying their meta-data.

Overlays the annotations onto the image.

Overlays object annotations from the report.

Ancestor for left-click processors that make use of SelectionRectangle.

Allows the user to create fixed-sized bounding boxes around the left-click position (<ctrl> left-click in box removes it).

Ancestor for selection processors that make use of the SelectionRectangle class.

Ancestor for point annotators.

Ancestor for annotators that use reports to store the annotations in.

For annotating object shapes with polygons boxes.

For annotating single points.

Flexible overlay for object annotations.