RodTracker

[v0.6.5]

Added

shows warning message when loading coordinate data for only one camera (instead of producing an error)
shows warning message when attempting to solve tracking or update the graphs when only one camera data is loaded

Fixed

updated documentation of RodTracker with better explanation of workflow and usage of example data
other small updates to documentation

[v0.6.4]

Added

missing sections in documentation were added (i.e. world coordinate transformation)
placeholders (TBDs) were replaced for API documentation

Changed

default rod detection models (CPU and CUDA) are now published and downloaded via PyTorch Hub (#102)
default number of detected particles changed to 25 for each color (corresponding to default sample data)
ParticleDetection dependency updated (ParticleDetection v0.4.3 now required)
reverted to original dialogue window for stereo calibration/world transformation (for consistency with image loading dialogue)

Fixed

tracking now works as intended, previously it required to press the “solve” button frame by frame (#96)
solved issue when changes in data were not saved and reverted after pressing “solve” button
tracking/matching tests were updated according to the fixed tracking issue

v0.6.3 

Added

support for Python 3.12
viewing mode displaying all particles in a frame (#94)
deleting particles from the Particles tab (#93)
informative dialog before downloading the example model for rod detection (#89)

Changed

removed scaling of example data (#95)
made error dialogues modal (#92)
folders must be selected during image loading (#88)

Fixed

numpy & torchvision version conflict (#90)
loss of dimension when detecting only 1 particle per frame (#91)
multiple errors in tests

v0.6.2 

Added

automated testing using a GitHub workflow (#84)
popup dialog showing unhandled errors to users (#74)
button to download an example detection model (#75, #85)
link to the source code in the documentation (#81)

Changed

removed/changed URL dependencies (#78)
removed metadata from version generation

Fixed

import sorting (#80)
outdated ParticleDetection dependency (#74)

Removed

GPU installation option

v0.6.1 

Added

support for python3.11
support for more complex image file names, i.e. ‘…_00000.png’
buttons to report issues/request features on GitHub
dynamic versioning
support for macOS
creation of installers on Windows, Linux, and macOS
choice between opening a local/online version of the documentation

Changed

logo/icon

Fixed

missing installation requirement (Git)
capability to ‘correct’ rods when none are present in the loaded dataset on the current frame (#73)

Removed

bundling into a single file

v0.6.0 

Added

old rod positions are displayed during editing as a visual cue
mode for automatic selection of the rod closest to the cursor
documentation with Sphinx
integration of ParticleDetection functionality
- detection of rods from images
- recalculation of 3D rod positions from position updates in 2D
- tracking of rods and following recalculation of 3D positions
- evaluation plots for 3D position reconstruction
shortcut to fit the displayed image to its available space
added a ‘busy’ indicator for tabs running long background tasks
shortcut to delete a rod when having it selected
display of the documentation from the Help dropdown menu
loaded/changed position data is automatically saved every 60 seconds

Changed

rods are now displayed partially transparent while being selected
display of a folder selection dialog, if no folder is selected during saving
settings-/data-directory is now determined with platformdirs

Fixed

test suite problems have been fixed
initial display of rods on Windows
rods not being completely deletable (#69)

Removed

function to display the README in the application
True number of rods setting is no longer available

v0.5.8 

Changed

migration from setuptools to poetry as the build system

Fixed

bundling of the RodTracker to a standalone program

v0.5.7 

Added

splash screen during the startup of the RodTracker

Changed

each module can now access its own logger, leading to more informative log messages

Fixed

cropping of displayed rod numbers for font sizes >11

Removed

splash screen during unpacking stage of standalone programs bundled as one file

v0.5.6

skipped

v0.5.5

skipped

v0.5.4 

Changed

3D rod elements are now reused instead of regenerated leading to better performance

Fixed

visual indicators of the experiment box are no longer extending further than the experiment’s dimensions

v0.5.3 

Changed

The rod data displayed as a tree is now updated in place instead of being regenerated. This especially improves performance and responsiveness of the RodTracker when working with large datasets.

v0.5.2 

Changed

rod position data is handled by a dedicated object
images datasets are handled by one dedicated object per camera view
rod data display as a tree is done by a custom QTreeWidget
zooming and panning in the 3D environment has improved speeds for the interaction

v0.5.1 

Added

Three different display modes for rods in 3D

Fixed

redundant settings updates are removed

v0.5.0 

Added

3D view of the rod data in relation to the experimental container
warnings of the Qt framework are now also logged to the RodTracker’s log file.

Changed

The right portion of the data display is now designed as tabs
Setting are now accessible in a tab on the right of the main area of the RodTracker, instead of a dialog from a dropdown menu. The changed settings are now applied immediately.
Mechanism for internal notification of successful saving
internal processes of data selection and loading are now seperated

Fixed

automatic rod activation during a camera change to where no rods are available
The unsaved changes indicator is now removed from all appropriate tabs, not only from the currently active one.
aborting of image selection
application crashes due to mishandled threads
cross platform path incompatibilities

v0.4

skipped

v0.3

skipped

v0.2

skipped

v0.1.1 

Fixed

crashes of the RodTracker when new rods are introduced during use (see #60)

v0.1.0 

Added

a versioning system

ParticleDetection

Unreleased 

[v0.4.3]

Fixed

tracking now works as intended (#96)
tracking/matching tests were updated according to the fixed tracking issue

v0.4.2 

Added

support for Python 3.12

Fixed

error mitigation for failed projection error binning
numpy version mismatch (#90)
minimum test coverage

v0.4.1 

Added

support for python3.11
support on macOS
project metadata
automated testing using a GitHub workflow (#84)
installation instructions to the README.md (#79)
dynamic versioning

Changed

removed/changed URL dependencies (#78)

Fixed

import sorting (#80)
multiple dependency constraints, otherwise breaking functionality
issue of Tkinter not finding a usable init.tcl on Windows

Removed

CPU & GPU extra

v0.4.0 

Added

documentation with Sphinx
- improved API documentation
- include examples
more basic functions are available instead of just their runner functions, e.g. project_points(...)
support for python 3.8 added
possibility to define an expected number of particles when extracting rod endpoints
added automated tests
exporting detected objects to JSON format for training
support for all annotation types provided by the VGG Image Annotator in training metadata
endpoint reordering for smoother post-processing

Changed

now uses the headless version of OpenCV
simplified the default format for transformations from camera to world coordinate system
object detection function now accepts arbitrary saving functions instead of having one fixed saving method built into it

Fixed

tracking of rods over multiple frames (see here)
incomplete/incorrect setup of logging for the library

Removed

run_detection(...) no longer returns the raw detection results

v0.3.1 

Added

start versioning