Plankton Identifier

Presentation

Plankton Identifier is a free software which allows the automatic identification of objects (planktonic organisms and/or particles) from a set of images with their associated data. Therefore, before you can use Plankton Identifier, you must extract object images (thumbnails in JPEG format) and object data (in PID or ZIM format) with a dedicated hardware/software combination such as ZooScan+ZooProcess, Flowcam+ZooProcess (see http://www.ZooScan.com ) or Flowcam+Zoo/PhytoImage ( http://www.sciviews.org/zooimage ).

Once data acquisition is done, Plankton Identifier helps the user to build a Learning File (set of objects identified by an expert) through a safe interface which prevents from data loss, erroneous duplicates etc. Then, using the learning file, Plankton Identifier allows the automatic recognition of an unlimited number of objects from one or several data files. Additionally, the user can optimize the recognition by many ways (by choice of different evaluation methods, selection of variables, variable transformation, way of grouping objects etc.). Automatic identification can be easily visualized to allow validation by experts. Validation results are stored in one or several text files which can easily be imported in other softwares (for graphics, time series analysis etc...). A detailed analysis report is also generated (html format allowing copy-paste).

NOTE: Although Plankton Identifier has been initially developed to identify Planktonic organisms, it can virtually identify any kind of objects.

Data analyses are not made by Plankton Identifier itself but by the free data mining software TANAGRA , in batch mode. This is the reason why you must install Tanagra 1.4.48 or above on your computer before using Plankton Identifier .

NOTE: Experimented users can define their own data analysis methods for Plankton Identifier using Tanagra tdm format.

 

Downloads

Version currently available is 1.3.4. It is a Win32 application running under Windows XP, Windows 98, Windows 2000 and Windows 7 (32 and 64 bits version). A screen resolution of at least 1024x768 (96 dpi) is highly recommended. There is no LINUX or MacOS version.

To install Plankton Identifier click on the link below and save the installer on your computer. Then, execute the installer and follow the instructions.

DOWNLOAD PLANKTON IDENTIFIER (4 Mo)

Eight recognition methods (all tested) are currently installed within Plankton Identifier as well as a method to test and compare the efficiency of each of them. Some demo files (zooplankton JPEG images and PID files) are included and will be installed in a 'Demo' subfolder.

IMPORTANT NOTE: If it is not already the case, you must set the decimal separator to ‘.' in your windows configuration (Start > Parameters > Control Panel > Regional Settings or Control Panel> Region and Language> Format tab > Additional Settings> ) and in any software used to prepare the data. The use of another decimal separator leads to failure during the data analysis process.

 

An additional application, PID Viewer , can be downloaded separately. PID viewer helps to visualize PID files information (header, data, and basic statistics). Only the header section is editable, in the sense that one can go in and make changes in the document and save it (the original document will be kept as backup file). Once PID viewer is installed on your computer, you just have to double-click on PID files (Sample or Learning) to easily explore and/or compare their contents, even from Plankton Identifier.

DOWNLOAD PID VIEWER (0.6 Mo)

NOTE1: Manual edition of a PID file is not recommended but if necessary you can use any usual text editor.

NOTE2: PID Viewer can also be used to visualize ZIM files produced by Zoo/PhytoImage ( http://www.sciviews.org/zooimage/ )

 

Please, don't forget to mention Plankton Identifier when you produce results, notably in scientific publications. You can refer to Plankton Identifier as follow:

Gasparini Stéphane and Antajan Elvire (2007-2013). PLANKTON IDENTIFIER: a software for automatic recognition of planktonic organisms. http://www.obs-vlfr.fr/~gaspari/Plankton_Identifier/index.php

 

Support

More information about the use of Plankton Identifier is available in the User Guide.

If you experience problems during installation, if you don't know how to make choices among options or if you don't know how to interpret the results, you can also look at the F.A.Q.

 

Known limitations

- Each PID file (sample) can contain an unlimited number of objects but it is not possible to analyse more than 392 PID files at once.

- All files (Sample and/or Learning) used in one analysis must have exactly the same variables (same order, same names). If you have files with different variables, use them in separate analysis.

- Analysis conducted with corrupted PID files can produce unexpected results. If you modify PID files manually, be sure you know what you are doing ! Look at the User guide for more information about PID files expected format.

 

Version history

Date
Version
Addition/Changes

15 Oct. 2013

1.3.4

New Plankton Identifier Release with major changes :

- Visual validation of prediction

- Compilation of results

- Separation of learning set performances evaluation and sample prediction in two different steps

- Win7 total compatibility

- User experience improvement

See maxime Ledée Blog here or new user guide here for more details.

26 July 2013
1.3
- Thanks to Maxime Ledee, a totally new Pid Viewer release (version 1.3) is now available, with edition ability and much more. An totally new version of PkID will also be released soon (see here for more detail)
6 May 2008
1.2.6

- Methods (tdm files) provided with PkID 1.2.6 updated to work with Tanagra version 1.4.22.

- New Pid Viewer release (version 1.1), with minor bugs fixed.

30 Mar.2007
1.2.6

- Comma and dot are now both supported as decimal separators. PID files decimal separator is still dot but you don't need to change Windows regional settings anymore to use Plankton Identifier and/or PID_viewer.

- New popup menu (right-click) with "Select all", "Invert Selection", "Delete" and "Move to..." habilities for sorted thumbnails (multiselect is now allowed).

- Improved files compatibility verification (to avoid mistake)

- Sound alert (needs a sound card) at the end of data analysis to wakeup the user when analysis process takes some time ;)

- 7 cross-validation methods (one for each supervised learning method) now available.

- Several minor bugs fixed.

22 Mar. 2007
1.2.5

- A bug impeding the use of modified group lists in some circumstances has been fixed.

- Change of defaut folder (settings) now update all the starting path.

- A new demo set (much bigger) has been included.

- A new method (cross-validation 1 (Rnd-Tree)) has been included.

14 Feb. 2007
1.2.4


- To avoid potential conflicts, several instances of Plankton Identifier can not be run simultaneously anymore.

- An animated window is now shown during data analysis to testify that Plankton Identifier is still running when analysis process takes some time.

- Some changes in demo files.

- Release of PID viewer 1.0 as separated application

15 Jan. 2007
1.2.3

- First public release

2006
1.0.0 to 1.2.2


- Beta versions

 

Sources

Plankton Identifier has been developed using IDE Borland Studio Developer 2006 (Delphi 10), which is free under conditions, and the free components Virtual Treeview (Soft-Gems) and VirtualShellTools (Mustangpeak). PID viewer uses the free Delphi component Toneinstance (delphipages.com). The installers have been created using Inno setup 5.5.3 (Jordan Russell's software). Plankton Identifier and PID viewer source codes are available on demand for non commercial purpose.

 

Acknowledgements

Thanks are due to Franck Coppin, Gaby Gorsky, Maxime Lédée, Marc Picheral and Ricco Rakotomalala for support and contribution to this project.

 

Some additional links

ImageJ : Free image analysis software (+ source code in Java)
ImageTools : Free image analysis sofware for Windows ( + source code in C)
NIH-image: Free image analysis software for MacOS (+source code in Pascal)
efg's computer lab: Technical informations about image processing, algorithms and much more.


Stéphane Gasparini

Laboratoire d'Océanographie de Villefranche

UMR 7093 - CNRS / Université Pierre et Marie Curie - Paris 6

BP 28 - 06234 Villefranche sur mer Cedex - France

Elvire Antajan

IFREMER

150, quai Gambetta

62200 Boulogne sur Mer - France


Last Update : 15-Oct-2013