blockedImageAutomationAlgorithm

Implements the blocked image automation algorithm

Since R2021a

Syntax

res = blockedImageAutomationAlgorithm(algObj,bstruct)

Description

res = blockedImageAutomationAlgorithm(algObj,bstruct) computes labels based on the algorithm you specify in this method. The Image Labeler app invokes this method on each image you choose for blocked image automation in the app. algObj is a vision.labeler.AutomationAlgorithm object. bstruct is a structure. The format of the output, res, depends on the type of automation algorithm specified.

Examples

Sample Blocked Image Algorithm Automation Function

• Here is an example of a blocked image algorithm automation function.

  function res = blockedImageAutomationAlgorithm(algObj,bstruct)
     % Detect people using aggregate channel features
     detector = peopleDetectorACF('inria-100x41');
     [bboxes, scores] = detect(detector, bstruct.Data,...
              'SelectStrongest', false);
             
     % Apply non-maximum suppression to select the strongest bounding boxes.
     [selectedBboxes, selectedScores] = selectStrongestBbox(bboxes, scores,...
              'RatioType', 'Min',...
              'OverlapThreshold', 0.65);
             
     % Consider only detections that meet specified score threshold
     selectedBboxes = selectedBboxes(selectedScores > 0, :);
     selectedBboxes(:,1) = selectedBboxes(:,1)+bstruct.Start(2);
     selectedBboxes(:,2) = selectedBboxes(:,2)+bstruct.Start(1);
             
     if algObj.SelectedLabelDefinitions.Type == "Rectangle"
        % Add the selected label at the bounding box position(s)
        res = struct(...
            'Type', labelType.Rectangle,...
            'Name', algObj.SelectedLabelDefinitions.Name,...
            'Position', selectedBboxes);
     end
  end

Input Arguments

algObj — Automation algorithm

vision.labeler.AutomationAlgorithm object

Automation algorithm, specified as a vision.labeler.AutomationAlgorithm object.

bstruct — Structure of data passed to function

scalar struct

Structure of data passed to the blocked image apply object function, specified as a scalar struct with these fields.

Field	Description
Data	A block of array data from the blocked image.
Start	Array subscripts of the first element in the block. If BorderSize is specified, this subscript can be out-of-bounds for edge blocks.
End	Array subscripts of the last element in the block. If BorderSize is specified, this subscript can be out-of-bounds for edge blocks.
Blocksub	The block subscripts of the current block
BorderSize	The value of the BorderSize parameter
BlockSize	The value of the BlockSize parameter. Note: size(data) can be less than this value for border blocks PadPartialValue is false.
BatchSize	The value of the BatchSize parameter

Output Arguments

res — Result of processing

scalar struct

Result of processing, returned as a scalar struct. The format of the structure depends on the type of automation algorithm.

For automation algorithms without pixel labels, the struct has fields described in this table.

Field	Description
Type	A labelType enumeration that defines the type of label. Type can have values: Rectangle, Line, Projected cuboid, Cuboid, or Scene.
Name	A character vector specifying a label name that returns true for checkLabelDefinition. Only existing label names previously defined in the Image Labeler app can be used.
Position	Positions of the labels. The type of label determines the format of the position data. Label Type Description Rectangle P-by-1 cell array specifying P Rectangles, each containing a 1-by-4 vector specifying position of the bounding box locations as [x y w h] or multiple Rectangle ROIs specified as an M-by-4 matrix. Line P-by-1 cell array specifying P polylines each containing N-by-2 vector specifying N points along each polyline as: [x1,y1; x2,y2;...xN,yN] Polygons P-by-1 cell array specifying P polygons each containing N-by-2 vector specifying N points along each polygon as: [x1,y1; x2,y2;...xN,yN] Projected Cuboid P-by-1 cell array specifying P projected cuboids each containing 1-by-8 vector specifying position of primary and secondary faces as: [x1 y1 w1 h1 x2 y2 w2 h2] or multiple projected cuboid ROIs can be specified as an M-by-8 matrix.
Attributes	An array of structs representing the attributes contained by the automated labels. Each attribute is specified as a field of the struct with the name of the field representing the name of the attribute and the value of the field representing the value of the attribute. This is an optional field, only present if the labels defined have attributes.

For automation algorithms with pixel labels, res is a categorical label matrix, where each category represents a pixel label.

Tips

• For automation algorithms without pixel labels, the Position field in res must be in a world coordinate system. This can be achieved by adding the X and Y indices in the bstruct.Start field to the output of the automation algorithm. To get the correct X coordinate, add bstruct.Start(2) to get the position of the automation output in world coordinates. To get the correct Y coordinate, add bstruct.Start(1) to get the position of the automation output in world coordinates.

Version History

Introduced in R2021a