Syntax

   FACTORY_DEF <ReaderKeyword> RasterObjectDetectFactory
   [FACTORY_NAME <factory name>]
   [INPUT FEATURE_TYPE <factory type>
      [<attribute name> <attribute value>]*
      [<feature function>]*]*
   [DETECTION_TYPE (FACE|BODY|ANIMAL|OBJECT)]
   [DET_SUB_TYPE_FACE <one or more space separated face detection type(s)>]
   [DET_SUB_TYPE_BODY <one or more space separated body detection type(s)>]
   [DET_SUB_TYPE_ANIMAL <one or more space separated animal detection type(s)>]
   [DET_SUB_TYPE_OBJECT <one or more space separated object detection type(s)>]
   [DET_SUB_TYPE_CUSTOM <path to the custom detection model .xml file>]
   [DEST_ATTR <string name of the detected object model attribute>]
   [SCALE_FACTOR <scale factor value 1-300>]
   [MIN_NUM_NEIGHBOR <min number of neighbors value>]
   [OBJ_SIZE_UNITS (Pixels|Percent)]
   [MIN_SIZE_WIDTH <minimum object size width>]
   [MIN_SIZE_HEIGHT <minimum object size height>]
   [MAX_SIZE_WIDTH <maximum object size width>]
   [MAX_SIZE_HEIGHT <maximum object size height>]
   [PRESERVE_ATTRIBUTES (Yes|No)]
   [OUTPUT DETECTED FEATURE_TYPE <feature type>
      [<attribute name> <attribute value>]*
      [<feature function>]*]*

Overview

   This factory is used to detect various objects on supplied rasters. Object detection models
   are provided by the OpenCV vision library and the list of available detection models are the
   default ones supplied with library source.

   Multiple detection models may be selected. The factory will output a feature with a rectangular
   (polygon) geometry representing the bounding box of a particular detected object. Each feature
   will be tagged with $(DEST_ATTR) attribute, the value of which will be equal to the
   name of the model that detected the object.

   Each detection type is essentially a mask or schema that is dragged across the entire raster and
   detected objects are pixel patterns that sufficiently match the expected schema. Because the schema
   is fixed, the image is scaled down in attempt to detect objects of all sizes.

   The speed of scaling can be configured with scaling factor. Minimum and maximum detection object
   sizes may be supplied to ignore all detected objects that are smaller or larger than the provided
   sizes. The sizes are in pixels, but a percent amount relative to the source raster can be specified

   If minimum size of the detected object is the same as its maximum size the image will not be
   scaled and detection will be performed at a single scale/size


Parameters

   DETECTION_TYPE

      [FACE|BODY|ANIMAL|OBJECT]. Determines the sub-group of the detection models

   DET_SUB_TYPE_FACE

      A multi-choice list of face detection models

   DET_SUB_TYPE_BODY

      A multi-choice list of body detection models

   DET_SUB_TYPE_ANIMAL

      A multi-choice list of animal detection models

   DET_SUB_TYPE_OBJECT

      A multi-choice list of general object detection models

   DET_SUB_TYPE_CUSTOM

      A file-select prompt for a custom detection model file (.xml)

   DEST_ATTR

      Determines the attribute name to be used for tagging detected objects with the detection model
      name that produced the object. Default: 'opencv_detection_type'

   SCALE_FACTOR <scale factor value 1-300>]

      Must be a value between 1 and 300, representing minimum of 1% and maximum of 300% scaling. The
      smaller the scaling the slower the detection model is going to run, however the results
      will be more granular. Larger scaling factor can be used for detecting large objects, in that
      case the models will run significantly faster

   MIN_NUM_NEIGHBOR

      The parameter specifies the minimum number of (likely overlapping) neighbors a detected object
      should have to be retained and be considered "detected". The parameter can be thought of as
      configuration for level of confidence in the object, to avoid some false positives

   OBJ_SIZE_UNITS

      Pixels or Percent. Determines if the units for the minimum and maximum sizes.
      - Pixels are self-explanatory pixel units
      - Percents are relative percentages of the source raster

   MIN_SIZE_WIDTH

      Minimum width of the object to search for

   MIN_SIZE_HEIGHT

      Minimum height of the object to search for

   MAX_SIZE_WIDTH

      Maximum width of the object to search for

   MAX_SIZE_HEIGHT

      Maximum height of the object to search for

   PRESERVE_ATTRIBUTES

      If the parameter is set to Yes, each output detected object feature will copy the attributes
      from the input raster

Output Tags

   The RasterConvolverFactory supports the following output tags:

   DETECTED

      Outputs a feature for each detected object, which is represented by a rectangular (polygon)
      bounding box. Each feature will contain an attribute, with a name determined from DEST_ATTR,
      parameter containing the name of the detection model that detected the object

   REJECTED

      If there are any incorrect transformer parameters or the input is incorrect, the feature with
      that raster will be rejected