Applications/MTBCellCounter: Difference between revisions

From MiToBo
Jump to navigationJump to search
(Created page with "<br/> <br/> == MTB Cell Counter == 150px|right|link= 150px|right|link= [[File:ActinDistro.png|150px|right|lin...")
 
 
(33 intermediate revisions by the same user not shown)
Line 3: Line 3:


== MTB Cell Counter ==
== MTB Cell Counter ==
[[File:ActinExample.png|150px|right|link=]]
[[File:MiToBoCellCounter-MainWindow.png|450px|right|link=]]
[[File:HT144-shC-Series010-clusters.png|150px|right|link=]]
[[File:ActinDistro.png|150px|right|link=]]


The ''MTB_CellCounter'' plugin is available since release version 1.5 of MiToBo.<br/>
The MTB Cell Counter plugin is dedicated to semi-automatic labeling and counting of small structures in images, like spots or cells. It is based upon the original Cell Counter plugin for ImageJ written by Kurt De Vos ([http://rsb.info.nih.gov/ij/plugins/cell-counter.html Cell Counter]) and now also available in [http://fiji.sc/Cell_Counter Fiji]. Compared to the original version the ''MTB_CellCounter'' plugin adds some nice new features:
 
* pre-segmentation and filtering of various structures
The plugin is based upon the original Cell Counter plugin for ImageJ written by Kurt De Vos ([http://rsb.info.nih.gov/ij/plugins/cell-counter.html Cell Counter]) and now available in [http://fiji.sc/Cell_Counter Fiji]. Compared to the original version the MTBCellCounter plugin adds some nice new features:
* pre-segmentation and filtering of spot-like structures
* free configuration of marker colors
* free configuration of marker colors
* advanced editing of markers
* advanced editing of markers
* status bar, tooltips and keyboard shortcuts  
* status bar, tooltips and keyboard shortcuts
<br>
 
=== News ===
In MiToBo 1.8.9 the plugin got an update. On the one hand its architecture was modified to support easy integration of additional segmentation operators, and on the other hand its usability was improved. Each type of marker can now individually be filtered for average intensity and size, and for new markers manually added by the user now also a region in terms of the surrounding contour can be specified.
 
Below you will find the updated documentation for the new version. Note that important new functionality is highlighted below.
 
Documentation for the old version (available in MiToBo 1.8.8 and earlier) is still available [[Applications/MTBCellCounter_1.0|here]].
<br>


===== Related Publications=====
=== Related Publications ===
* ''The Authors'','''"The 'MTB Cell Counter' - a versatile tool for semiautomated quantification of sub-cellular phenotypes in fluo-
* ''L. Franke, B. Storbeck, J. L. Erickson, D. Rödel, D. Schröter, B. Möller, and M. H. Schattat''. ''' ''The 'MTB Cell Counter' a versatile tool for the semi-automated quantification of sub-cellular phenotypes in fluorescence microscopy images. A case study on plastids, nuclei and peroxisomes.'' ''' Journal of Endocytobiosis and Cell Research, 26:31-42, 2015, [http://zs.thulb.uni-jena.de/receive/jportal_jparticle_00342296 Online version].
rescence microscopy images"'''.<br/>2015, submitted for publication.
<br>


===== Name of Plugin/Operator =====
=== Name of MiToBo Plugin/Operator ===
<code>
<code>
mtb_cellcounter.MTB_CellCounter
mtb_cellcounter.MTB_CellCounter
</code>
</code>
<br>
<br>
(available since MiToBo version 1.5)
(version 1.0 is available in MiToBo-Plugins since version 1.5, an updated version was released in MiToBo/MiToBo-Plugins 1.8.9)
<br>
<br>
<br>
=== Installation and Usage ===
===== Download and Installation =====
The easiest way to start working with the plugin is to install it in Fiji via MiToBo's update site.<br>
From the Fiji menu select ''"Help"'' -> ''"Update..."'', then wait until the updater checked the status of your installation. Afterwards select ''"Manage update sites"'' from the updater window and select ''"MiToBo"''. Close and cancel all windows and restart Fiji. After the restart you will find the MTBCellCounter via ''"Plugins"'' -> ''"MiToBo"'', and finally ''"MTBCellCounter"''.
For a quick start in ImageJ (not Fiji) you best download the MiToBo-Plugins zip file from [https://moon.informatik.uni-halle.de/archiva/repository/releases/de/unihalle/informatik/MiToBo/mitobo-plugins/1.8.9/mitobo-plugins-1.8.9-bin.zip here]. On Linux and Mac OS systems just unzip the file in an empty directory of your choice and start ImageJ by running the "run.sh" script included in the zip file.
On Windows machines it is easiest to first install ImageJ or Fiji and then follow the instructions given [http://www2.informatik.uni-halle.de/agprbio/mitobo/index.php/Installation#Windows here].
Further information on how to install MiToBo in general can be found on the [http://www2.informatik.uni-halle.de/agprbio/mitobo/index.php/Installation Installation page].
Once MiToBo is installed, you can start the MTB_CellCounter from the ''"Plugins"'' menu, item ''"MiToBo"'', then ''"MTB CellCounter"''.


===== Usage =====
===== Usage =====
For using the plugin you need to install MiToBo by following the instructions on the [[Installation]] page. Running ImageJ you will then find a new entry 'MiToBo' in the plugins menu from where you can select the 'MTB CellCounter' plugin.
The usage of the plugin is leaned on the usage of the original CellCounter plugin (see also [http://rsb.info.nih.gov/ij/plugins/cell-counter.html Cell Counter webpage]), but extended with additional preprocessing steps to pre-segment objects of interest in the image.
 
The usage of the plugin is leaned on the usage of the original plugin (see also [http://rsb.info.nih.gov/ij/plugins/cell-counter.html Cell Counter webpage]).


The basic workflow is as follows:
The basic workflow is as follows:
* open the image you would like to process and press the 'Initialize' button
* open the image you would like to process and press the ''"Initialize"'' button
* optionally configure the particle detector via the 'Configure operator...' button and then press 'Detect'
* optionally '''select one or many of the available object detectors, configure each detector individually, and then press ''"Run Detectors"'''''
* once the detection is finished you can filter detected particles via the 'Filter Particles...' button by size and average intensity; if you are done, press 'Select Markers'
* once the detection is finished you can filter detected objects by size and average intensity '''for each marker type separately''' using the ''"Filter Objects of Active Type..."'' button;<br> if you are done, press ''"Select Markers"''
* now markers can manually be postprocessed, i.e. markers can be added or removed, or their type can be changed
* now markers can manually be edited, i.e. markers '''(and corresponding contours)''' can be added or removed, or their type can be changed
* at the end you can view marker statistics (button 'Results'), save the markers to a file (button 'Save Markers') or do some measurements (button 'Measurements...')
* at the end you can view marker statistics (button ''"Results"''), save the markers to a file (button ''"Save Markers"'') or do some measurements (button ''"Measurements..."'')


===== Functions and Options =====
===== Functions and Options =====
Below we outline the functions of the various elements of the graphical user interface.
Below we outline the functions of the various elements of the graphical user interface.


<br><br>
* '''Initialization:'''
* Initialization:
** ''"Initialize"'': initializes the plugin with the currently active image
** 'Initialize': initializes the plugin with the active image, if the image has more than one channel only the first channel is considered
** ''"Keep original"'': if checked the source image remains open, otherwise it is closed
** 'Keep original': if checked the source image remains open, otherwise it is closed
 
* '''Configuration of input image:'''
** '''''"Channel to process"''''': here you can select on which channel your are going to work if your image has more than one channel; note that you can freely switch between the channels while labeling your objects
** '''''"Boundary Channel (opt)"''''': for labeling sub-cellular objects it is possible to overlay the image with cell contours for easier orientation;<br>the contours have to be provided in binary format in one of the channels of the input image where the background should have zero intensity and the boundaries be labeled with a unique intensity or color value greater than zero


* Pre-segmentation:
[[File:ScreenshotParticleFilter.png|250px|right|link=]]
** 'Detect': runs the particle detector
* '''Pre-segmentation:'''
** 'Configure Operator...': allows to change the parameters of the particle detector
** ''"Add detector"'': will add the selected detector to the set of "Selected detectors"; you can select a detector from the list of "Available object detectors" by clicking on its entry
** 'Filter Particles...': allows to filter detection results
** ''"Remove"'': will remove the selected detector from the list of "Selected detectors"; you can select a detector from the list of "Selected detectors" by clicking on its entry
** 'Show contours': enables/disables display of the contours of detected particle regions
** ''"Configure..."'': allows to change the parameters of the detector currently selected (see below)
** 'Select markers': selects the final set of markers and terminates detection stage
** ''"Run Detectors"'': will apply all selected detectors to the currently selected channel of the input image
** ''"Filter Objects of Active Type..."'': allows to filter detection results of the currently selected type for size and average intensity
** ''"Show contours"'': enables/disables display of the contours of detected particle regions
** ''"Select Markers"'': selects the final set of markers (after running detectors and filtering results) and terminates detection stage


Detected particles are labeled with marker type 1 and the counter of that type refers to their number.
Note that for each "Selected detector" you need to specify a marker type, i.e. an index between 1 and the number of available counters on the left.<br>
Objects detected by the corresponding detector will be labeled with the selected type.<br>
In case that there exist already markers of that type, you will get a warning that all existing markers will get lost on running the detector.


* Manual post-processing:
* '''Marker management and editing:'''
** 'Add': add a new marker type at the end of the list, the type gets a random color
** ''"Add"'': adds a new marker type (also named "counter") to the end of the list, the new counter gets a random color
** 'Remove': delete the last type from the end of the list
** ''"Remove"'': deletes the last counter (and all of its markers) from the end of the list
** 'Delete': delete the last placed marker
** ''"Delete"'': deletes the last placed marker
** ''"Reset"'': deletes all markers
** ''"Show Markers"'': enables/disables display of markers
** ''"Show Numbers"'': enables/disables display of marker numbers
** ''"Show All"'': enables/disables display of both markers and numbers


Note that the currently selected marker type and the settings for showing markers and numbers are also displayed in the status bar of the image window.<br>
Some additional actions for editing markers are available via keyboard shortcuts and mouse actions only, see below.


* '''Results:'''
** ''"Results"'': shows table with marker statistics
** ''"Save Markers"'': saves the markers to an XML file
** ''"Load Markers"'': loads markers from an XML file
** ''"Export Image"'': save a copy of the image including all markers


* Parameters:
===== Keyboard Shortcuts and Mouse Actions =====
{|class="wikitable"
{|class="wikitable"
|Name
|Key
|Description
|Description
|-
|-
|''Image directory''
|''1-9''
|directory where the input image data can be found
|select the corresponding marker type
|-
|''Mask directory''
|directory where the label images or region boundary files can be found
|-
|''Mask format''
|format of the segmentation data files: LABEL_IMAGE = images with unique labels for each cell and a value of zero for the background / IJ_ROIS = set of ImageJ 1.x ROIs, one ROI for each cell
|-
|''Output and working directory''
|directory to which the result files and intermediate data is written
|-
|-
|''Calculate features''
|''a'' or ''left arrow''
|if disabled the operator expects the features to be already present in the input directory and skips the (time-consuming) feature calculations;<br> this option is helpful if the features have already been calculated ones and only the parameters of the clustering should be changed
|scroll image to the left
|-
|-
|''Feature directory''
|''d'' or ''right arrow''
|directory where the features should be saved or - in case they are already available - from where they are read; the directory can be the same as the output and working directory
|scroll image to the right
|-
|-
|''Tile size x/y''
|''e''
|size of the sliding window used for feature calculations, should be chosen according to the resolution of the input images
|zoom out
|-
|-
|''Tile shift x/y''
|''q''
|shift of the sliding window, if the shift is smaller than the tile size sliding windows overlap
|zoom in
|-
|-
|''Distance''
|''s'' or ''arrow down''
|pixel-pair distance in co-occurence matrix calculations
|scroll downwards
|-
|-
|''Set of directions''
|''w'' or ''arrow up''
|directions to be considered in co-occurence matrix calculations
|scroll upwards
|-
|-
|''Isotropic calculations''
|''v''
|the texture features are derived from co-occurence matrixes; if this flag is enabled features for different directions are averaged,<br> otherwise all individual directions are preserved (resulting in larger, but also more informative feature vectors)
|enable/disable display of contours (only after detected markers are selected)
|-
|-
|''Number of feature clusters''
|''x''
|number of clusters in first stage, i.e., number of expected structural patterns in the images
|enable/disable display of numbers
|-
|-
|''Do PCA in stage II?''
|''y''
|allows to enable/disable PCA on the cluster distribution vectors prior to the pairwise distance calculations; by default enabled
|enable/disable display of markers
|}
|}


===== Additional Tools =====
In edit mode the following mouse actions are available:
The hierarchical clustering in stage II of our approach as described in the paper has been done using the [http://deim.urv.cat/~sgomez/multidendrograms.php MultiDendrograms] software.<br> In principal every hierarchical clustering tool can be applied. <br>
* ''left mouse button'': place a new marker of currently selected type
The basis for the hierarchical cluster analysis is the file ''AllImagesPairwiseDistanceData.txt'' to be found in the output directory upon termination of an analysis run. It contains a matrix of pairwise Euclidean distances of the (optionally dimension-reduced) cluster distribution vectors of all cells. The file can directly be loaded by MultiDendrograms, for other tools format convertion might be necessary.<br>
* ''right mouse button'': delete nearest marker of currently selected type
* ''Strg'' + ''left mouse button'': change type of nearest marker (with a different type) to currently selected type
* ''Shift'' + ''left mouse button'': draw a closed region of interest, all markers inside the region are removed
* '''''Shift'' + ''Ctrl'' + ''left mouse button'': add a new marker (for a region) by drawing its contour'''
<br>


You can download the latest version of MultiDendrograms from its webpage: [http://deim.urv.cat/~sgomez/multidendrograms.php]
=== Available Detectors ===
For detecting structures in the image the plugin provides a set of detectors which are described in more detail below.
==== Particles with UWT ====
[[File:MiToBoCellCounter-UWTDetector.png|350px|right|link=]]
For the pre-segmentation of spot-like structures a particle detector operator available in the core of MiToBo is used. It has been published in<br>


===== Sample data =====
''O. Greß, B. Möller, N. Stöhr, S. Hüttelmaier and S. Posch'',<br>
For testing the ''ActinAnalyzer2D'' operator we provide some sample data:
''' ''Scale-adaptive Wavelet-based Particle Detection in Microscopy Images,'' '''<br>
[http://www.informatik.uni-halle.de/mitobo/downloads/actin_examples.zip ActinAnalyzer2D sample data]
Proc. of Workshop Bildverarbeitung für die Medizin (BVM '10), Hans-Peter Meinzer, Thomas Martin Deserno, Heinz Handels, and Thomas Tolxdorff, editors,<br>
Springer, Informatik Aktuell, pp. 266-270, Aachen, Germany, March 2010.
<br>


The archive contains the following sub-folders:
The operator targets at spot-like structures over multiple scales. It basically relies on an undecimated wavelet transformation (UWT) combined with a probabilistic framework to determine for each detection event the optimal scale.
* ''imageData'': test images that were used in the ICPR publication mentioned above
The operator offers the following parameters:
* ''maskData'': corresponding label images
{|class="wikitable"
* ''featureData'': pre-calculated features for the cells (re-calculating the features may require up to an hour, depending on the machine used)
|Name
* ''resultData'': sample results calculated on the given data
|Description
 
|-
In addition, in the archive a file with a sample parameter configuration for the operator can be found. The parameters are those used for producing the sample results. Once the operator has been started the file can be loaded via the '' 'File' '' menu and its entry '' 'Load Settings' ''. Note that you need to set the image and mask directories, and also the feature directory according to your local file system structure and the place to where you extracted the zip file.
|''Minimal Scale (JMin)''
 
|scale of smallest objects, set to an integer value >= 1
For more information on the data and the morphological analysis of the cells, see<br> <br>
|-
''Anne Zirkel, Marcell Lederer, Nadine Stöhr, Nikolaos Pazaitis, and Stefan Hüttelmaier''<br>
|''Maximal Scale (JMax)''
'''''IGF2BP1 promotes mesenchymal cell properties and migration of tumor-derived cells by enhancing the expression of LEF1 and SNAI2 (SLUG)'''''<br>
|scale of largest objects, set to an integer value > JMin
''Nucleic Acids Res. Jul 2013; 41(13): 6618–6636. Published online May 15, 2013. doi: 10.1093/nar/gkt410'', [http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3711427/ Article]
|-
|''Scale Interval Size''
|several adjacent scales are correlated, the interval size determines how many scales are considered for each correlation
|-
|''Correlation Threshold''
|threshold for detecting objects, the smaller the threshold is chosen the more objects will be detected
|-
|''Minimum Region Size''
|minimal size of valid objects (in pixels), smaller objects are discarded
|}
<br>


=== Updates ===
=== Sample data ===
----<br>
For testing you can use the following set of sample images:


'''July 2014'''
* [http://www2.informatik.uni-halle.de/mitobo/downloads/CellCounter-SampleImages.zip Cell counter sample images]
* Released first version of Actin Analyzer as published in ICPR 2014.

Latest revision as of 14:13, 25 April 2018



MTB Cell Counter

MiToBoCellCounter-MainWindow.png

The MTB Cell Counter plugin is dedicated to semi-automatic labeling and counting of small structures in images, like spots or cells. It is based upon the original Cell Counter plugin for ImageJ written by Kurt De Vos (Cell Counter) and now also available in Fiji. Compared to the original version the MTB_CellCounter plugin adds some nice new features:

  • pre-segmentation and filtering of various structures
  • free configuration of marker colors
  • advanced editing of markers
  • status bar, tooltips and keyboard shortcuts


News

In MiToBo 1.8.9 the plugin got an update. On the one hand its architecture was modified to support easy integration of additional segmentation operators, and on the other hand its usability was improved. Each type of marker can now individually be filtered for average intensity and size, and for new markers manually added by the user now also a region in terms of the surrounding contour can be specified.

Below you will find the updated documentation for the new version. Note that important new functionality is highlighted below.

Documentation for the old version (available in MiToBo 1.8.8 and earlier) is still available here.

Related Publications

  • L. Franke, B. Storbeck, J. L. Erickson, D. Rödel, D. Schröter, B. Möller, and M. H. Schattat. The 'MTB Cell Counter' a versatile tool for the semi-automated quantification of sub-cellular phenotypes in fluorescence microscopy images. A case study on plastids, nuclei and peroxisomes. Journal of Endocytobiosis and Cell Research, 26:31-42, 2015, Online version.


Name of MiToBo Plugin/Operator

mtb_cellcounter.MTB_CellCounter
(version 1.0 is available in MiToBo-Plugins since version 1.5, an updated version was released in MiToBo/MiToBo-Plugins 1.8.9)

Installation and Usage

Download and Installation

The easiest way to start working with the plugin is to install it in Fiji via MiToBo's update site.
From the Fiji menu select "Help" -> "Update...", then wait until the updater checked the status of your installation. Afterwards select "Manage update sites" from the updater window and select "MiToBo". Close and cancel all windows and restart Fiji. After the restart you will find the MTBCellCounter via "Plugins" -> "MiToBo", and finally "MTBCellCounter".

For a quick start in ImageJ (not Fiji) you best download the MiToBo-Plugins zip file from here. On Linux and Mac OS systems just unzip the file in an empty directory of your choice and start ImageJ by running the "run.sh" script included in the zip file. On Windows machines it is easiest to first install ImageJ or Fiji and then follow the instructions given here.

Further information on how to install MiToBo in general can be found on the Installation page.

Once MiToBo is installed, you can start the MTB_CellCounter from the "Plugins" menu, item "MiToBo", then "MTB CellCounter".

Usage

The usage of the plugin is leaned on the usage of the original CellCounter plugin (see also Cell Counter webpage), but extended with additional preprocessing steps to pre-segment objects of interest in the image.

The basic workflow is as follows:

  • open the image you would like to process and press the "Initialize" button
  • optionally select one or many of the available object detectors, configure each detector individually, and then press "Run Detectors"
  • once the detection is finished you can filter detected objects by size and average intensity for each marker type separately using the "Filter Objects of Active Type..." button;
    if you are done, press "Select Markers"
  • now markers can manually be edited, i.e. markers (and corresponding contours) can be added or removed, or their type can be changed
  • at the end you can view marker statistics (button "Results"), save the markers to a file (button "Save Markers") or do some measurements (button "Measurements...")
Functions and Options

Below we outline the functions of the various elements of the graphical user interface.

  • Initialization:
    • "Initialize": initializes the plugin with the currently active image
    • "Keep original": if checked the source image remains open, otherwise it is closed
  • Configuration of input image:
    • "Channel to process": here you can select on which channel your are going to work if your image has more than one channel; note that you can freely switch between the channels while labeling your objects
    • "Boundary Channel (opt)": for labeling sub-cellular objects it is possible to overlay the image with cell contours for easier orientation;
      the contours have to be provided in binary format in one of the channels of the input image where the background should have zero intensity and the boundaries be labeled with a unique intensity or color value greater than zero
ScreenshotParticleFilter.png
  • Pre-segmentation:
    • "Add detector": will add the selected detector to the set of "Selected detectors"; you can select a detector from the list of "Available object detectors" by clicking on its entry
    • "Remove": will remove the selected detector from the list of "Selected detectors"; you can select a detector from the list of "Selected detectors" by clicking on its entry
    • "Configure...": allows to change the parameters of the detector currently selected (see below)
    • "Run Detectors": will apply all selected detectors to the currently selected channel of the input image
    • "Filter Objects of Active Type...": allows to filter detection results of the currently selected type for size and average intensity
    • "Show contours": enables/disables display of the contours of detected particle regions
    • "Select Markers": selects the final set of markers (after running detectors and filtering results) and terminates detection stage

Note that for each "Selected detector" you need to specify a marker type, i.e. an index between 1 and the number of available counters on the left.
Objects detected by the corresponding detector will be labeled with the selected type.
In case that there exist already markers of that type, you will get a warning that all existing markers will get lost on running the detector.

  • Marker management and editing:
    • "Add": adds a new marker type (also named "counter") to the end of the list, the new counter gets a random color
    • "Remove": deletes the last counter (and all of its markers) from the end of the list
    • "Delete": deletes the last placed marker
    • "Reset": deletes all markers
    • "Show Markers": enables/disables display of markers
    • "Show Numbers": enables/disables display of marker numbers
    • "Show All": enables/disables display of both markers and numbers

Note that the currently selected marker type and the settings for showing markers and numbers are also displayed in the status bar of the image window.
Some additional actions for editing markers are available via keyboard shortcuts and mouse actions only, see below.

  • Results:
    • "Results": shows table with marker statistics
    • "Save Markers": saves the markers to an XML file
    • "Load Markers": loads markers from an XML file
    • "Export Image": save a copy of the image including all markers
Keyboard Shortcuts and Mouse Actions
Key Description
1-9 select the corresponding marker type
a or left arrow scroll image to the left
d or right arrow scroll image to the right
e zoom out
q zoom in
s or arrow down scroll downwards
w or arrow up scroll upwards
v enable/disable display of contours (only after detected markers are selected)
x enable/disable display of numbers
y enable/disable display of markers

In edit mode the following mouse actions are available:

  • left mouse button: place a new marker of currently selected type
  • right mouse button: delete nearest marker of currently selected type
  • Strg + left mouse button: change type of nearest marker (with a different type) to currently selected type
  • Shift + left mouse button: draw a closed region of interest, all markers inside the region are removed
  • Shift + Ctrl + left mouse button: add a new marker (for a region) by drawing its contour


Available Detectors

For detecting structures in the image the plugin provides a set of detectors which are described in more detail below.

Particles with UWT

MiToBoCellCounter-UWTDetector.png

For the pre-segmentation of spot-like structures a particle detector operator available in the core of MiToBo is used. It has been published in

O. Greß, B. Möller, N. Stöhr, S. Hüttelmaier and S. Posch,
Scale-adaptive Wavelet-based Particle Detection in Microscopy Images,
Proc. of Workshop Bildverarbeitung für die Medizin (BVM '10), Hans-Peter Meinzer, Thomas Martin Deserno, Heinz Handels, and Thomas Tolxdorff, editors,
Springer, Informatik Aktuell, pp. 266-270, Aachen, Germany, March 2010.

The operator targets at spot-like structures over multiple scales. It basically relies on an undecimated wavelet transformation (UWT) combined with a probabilistic framework to determine for each detection event the optimal scale. The operator offers the following parameters:

Name Description
Minimal Scale (JMin) scale of smallest objects, set to an integer value >= 1
Maximal Scale (JMax) scale of largest objects, set to an integer value > JMin
Scale Interval Size several adjacent scales are correlated, the interval size determines how many scales are considered for each correlation
Correlation Threshold threshold for detecting objects, the smaller the threshold is chosen the more objects will be detected
Minimum Region Size minimal size of valid objects (in pixels), smaller objects are discarded


Sample data

For testing you can use the following set of sample images: