New Features - Bug Fixes and Limitations

New Features

New Features in eCognition 9.5.1:

Release 9.5.1 includes a list of bug fixes, whereas no new features are added to the software.

New Features in eCognition 9.5.0:

Story Feature Description
Point cloud analysis Improved algorithm 'rasterize point cloud' Performance improved by restricting analysis to the bounding box of the point cloud
Point cloud analysis Improved algorithm 'create temporary point cloud' A new 'downsampling' parameter allows for removal of redundant (dense) points
Point cloud analysis Improved algorithm 'assign class to point cloud' Multiple classes can be assigned at once (faster overall performance)
Point cloud analysis Improved performance Point clouds are not automatically rasterized
Point cloud analysis New algorithm 'remove points from point cloud' Algorithm removes points in an existing point cloud
Point cloud analysis Improved algorithm 'create temporary point cloud' Several subsets can be created at once (faster overall processing time)
Point cloud analysis Improved algorithm 'compute statistical value from point cloud' Several values can be computed at once (faster overall processing time)
Point cloud analysis Improved algorithm 'rasterize point cloud' Algorithm can now do several rasterizations at once (faster overall performance)
Point cloud analysis New point cloud scene features Scene features evaluate extreme coordinates and number of points in a point cloud
3D viewer Improved 3D viewer 2D images can be visualized in 3D viewer at user-defined elevation: View Settings Image Layer parameter Elevation (3D view)
3D viewer Improved 3D viewer Image object information window and attribute tables now display properties of vectors selected in 3D viewer
3D viewer Improved 3D viewer Area zoom now available in 3D viewer as well
Deep learning New algorithm 'delete sample space' Sample folders can be removed by executing an algorithm (to free disk space)
Deep learning New algorithm 'split labeled sample space' Samples in one folder can be divided into different independent sample folders
Deep learning New algorithm 'merge labeled sample space' Samples divided across different sample folders can be combined into a single sample folder
Deep learning Improved algorithm 'create convolutional neural network' A new option allows users to create models with batch normalization
Vector processing Improved algorithm 'create temporary thematic vector layer' Empty layers of various types can now be created
Vector processing New feature 'Classified as (at vector)' This feature allows to test for the existence of an image object class at the location of a vector
Vector processing Improved algorithm 'vector buffering' The buffer size can now also be defined in pixel units
Vector processing Improved algorithm 'create thematic vector object' The coordinates of the new vectors can now also be defined in pixel units
Vector processing Improved algorithm 'vector buffering' Algorithm is now called 'vector buffering/shrinking' and works also with negative buffer sizes
Vector processing New feature 'Overlap with parent process vector' This feature tests whether a domain vector overlaps the vector in the domain of the parent process
Vector processing Improved algorithm 'update region' Option 'from domain' sets the region to exactly include a vector or image object domain
Vector processing Improved vector algorithms Vector processing can be restricted to a region (faster processing)
Vector processing New feature 'Layer pixel value at point vector' This feature returns the layer pixel value of a user-defined layer at the location of a point vector
Vector processing New vector feature 'Z coordinate' This feature returns the z-coordinate of vectors
Vector processing Improved algorithm 'vector boolean' Domain 'current vector' is now enabled
Vector processing Improved algorithm 'delete vector objects' Domain 'current vector' is now enabled
Vector processing Improved algorithm 'compute statistical value' Vector domain is enabled to allow for computation of vector statistics
Basic algorithms Improved algorithm 'delete layer' New option to remove all temporary layers
Basic algorithms Improved algorithm 'delete image object level' New option to delete all image object levels
Basic algorithms Improved algorithm 'index layer calculation' New index option 'EVI (enhanced vegetation)' Enhanced Vegetation Index
Basic algorithms Improved algorithm 'index layer calculation' New index option 'NDSnI (snow)' Normalized Difference Snow Index
Basic algorithms Improved algorithm 'index layer calculation' New index option 'SAVI (soil-adjusted vegetation)' Soil Adjusted Vegetation Index
Basic algorithms Improved algorithm 'index layer calculation' New index option 'GRVI (green-red vegetation)' Green-Red Vegetation Index
Basic algorithms Improved algorithm 'template matching' New option to limit processing to a region
Basic algorithms Improved algorithm 'template matching' Templates in folder do not need to have the same size
Basic algorithms Improved algorithm 'template matching' Template image formats other than .tif enabled
Basic algorithms Improved algorithm 'template matching' Generated thematic layer with matches contains additional confidence attribute
Basic algorithms Improved algorithm 'watershed segmentation' Old image object level is overwritten automatically and objects are now connected by default
Table of content Improved window 'view settings' Different vector layer types are now identified by distinct icons
Table of content Improved perspectives Source view window is displayed in the default perspective
Table of content Improved window 'view settings' Layer alias renaming can be initiated with a double-click
Table of content Improved window 'view settings' Layer alias can be renamed via context menu
Table of content Improved window 'view settings' Layers can now be reordered in 'view settings' view
Table of content Improved usability .dpr files can be added to a workspace via drag & drop
Usability Improved warning User is warned to transform images in GCS to a projected data format
Usability Improved message Warn users that image object level is deleted when an image with different resolution is loaded and allow opt out
Usability Improved warning User is warned when loading an image with different x and y resolution
Usability Improved error message Errors during project saving indicate the source of the problem more clearly
Usability Improved error message User is warned with an explicit message when a vector layer is added to the project before it contains image data
Usability Improved text message Improved text message when user needs to decide between opening second viewer pane or 3D viewer
Application development New file shortcut The rule set can now reference {:Install.Dir} to refer to the bin folder of the software installation
Application development Improved parameter sets Parameter sets can now include double and string arrays
Application development Improved algorithm 'select input mode' Support for point cloud points and relative paths

Bug Fixes

The update includes the usual bug fixes and improved performance and system stability.

Bug Fixes in 9.5.1:

Reference Description
ECOG-5952 eCognition Engine Server doesn't start in grid installation on Windows due to missing .dll
ECOG-5976 eCognitionEarthServer64.9.5.0.15.dpk installation error causes memory overflow
ECOG-5990 Template matching samples and templates not displayed correctly in Template Editor when working on a subset
ECOG-6024 Fatal error in TensorFlow due to NVIDIA RTX and GTX graphics card series
ECOG-6038 Export image (*tif) algorithm doesn't work due to missing coordinate system information
ECOG-6044 Adjust default sample size settings in the create convolutional neural network algorithm
ECOG-6053 Flexera FlexNet Publisher vulnerability due to outdated FNP version
ECOG-6063 Shuffle labeled sample patches algorithm crashes when run on merged sample space
ECOG-6138 Crash in Image Object View due to invalid channel for haralick feature
ECOG-6162 Thematic raster file export RAT type incorrect due to GDAL code
ECOG-6168 Delete Level button unresponsive when 3D View window active
ECOG-6172 Layer alias duplicated after median filter algorithm executed

Bug Fixes in 9.5.0:

Reference Description
ECOG-4812 Developer does not start when temp file is set to a non-accessible network path
ECOG-5399 Aliases changed in 'manage layer alias' dialog are not updated in source view window
ECOG-5400 Process tree keyboard shortcuts don't work after first starting Architect and then Developer
ECOG-5401 Docking bar is not activated when toggling it if it is tabbed with other docking bars
ECOG-5427 Standard deviation split segmentation only works on main map
ECOG-5430 Source view window lists temporary vector layers as temporary raster layers
ECOG-5441 Index layer calculation algorithm wrong for NDWI
ECOG-5444 Channel statistics calculated for raster layer even when not displayed
ECOG-5445 Region filter in point cloud algorithms includes neighboring pixels
ECOG-5448 Changes in layer aliases via source view are not transferred to generated subsets
ECOG-5455 Software crashes when altering domain in customized algorithm
ECOG-5456 Customized algorithm domain is not transferred correctly
ECOG-5462 Crash in create/modify project dialog when changing a layer alias of a workspace project
ECOG-5489 Developer crashes during rule set editing with drop-down menus
ECOG-5510 User adjusted units in conditions sometimes unexpectedly go back to default
ECOG-5512 Point cloud region parameter does not work in algorithm 'create temporary point cloud' with modified fields
ECOG-5529 Not all points are rendered (tiling artifacts)
ECOG-5547 3D View shows incorrect intensity values of points
ECOG-5550 Algorithm 'template generation' does not work properly when 'template name' parameter is defined by a variable or a feature
ECOG-5554 Template matching returns vectors not clearly associated to a pixel (rounding issue)
ECOG-5556 Crash in image object info when selecting pixel
ECOG-5570 The t-dimension start/stop button in the 3D toolbar is not working correctly
ECOG-5581 Vector-based segmentation does not punch through all vectors
ECOG-5585 3 point 3D subset selection does not work when user clicks outside the scene boundary
ECOG-5595 Vectors that are larger than the current map cannot be included in the vector domain
ECOG-5603 Distance map algorithm does not adhere to compatibility settings correctly
ECOG-5611 Template matching creates a layer that may contain artifacts in homogeneous image regions
ECOG-5612 Uninstalling eCognition software leads to an error message
ECOG-5615 Coordinate of vector point is not consistent with coordinates of polygon
ECOG-5624 Load rule set sometimes ignores case sensitivity of class names
ECOG-5628 Adjustment of feature view fields freezes Developer
ECOG-5629 Software unexpectedly disappears when GPU version of deep learning is used
ECOG-5636 Erroneous mainline creation (for segment touching node)
ECOG-5681 Using a feature for 'number of cycles' causes problems during process editing
ECOG-5682 After editing a customized feature a subsequent feature is unexpectedly selected
ECOG-5683 Customized feature group (vector vs object feature) is not updated after editing
ECOG-5701 Performance of vector boolean affected by open view settings docking bar
ECOG-5702 Software sometimes crashes when updating the range in the feature view
ECOG-5718 Developer crashes with very large (40GB) .las import
ECOG-5757 Vector intersection sometimes give inaccurate result
ECOG-5764 Algorithm layer arithmetics interprets layers as scene variables (when there is a scene variable with the same name)

Known Issues and Limitations

The most important fix was made to the v9.5.1 License Server - it has been updated to account for security vulnerabilities. Therefore, users must update their License Server before they can use v9.5.1 client software packages. Please note, this will require users to return all active licenses before uninstalling the v9.5.0 License Server.

Prior to installing the eCognition License Server version 9.5.1 users should return all licenses and uninstall previous versions of the License Server. All 9.5.1 products require the eCognition License Server version 9.5.1 to work.

The eCognition License Server does not support the web interface on Linux distribution CentOS 6 (license server startup script needs to be modified to use -noWeb option for lmadmin).

Please contact support if you have further questions.