22.10.23:
- Refactoring the UI, reorganizing the tabs, set the main tab "Thematic map" in
the top
always visible. Changes in the layout for selecting the layer and its band and
nodata in
one row, reducing the space and improving the usability. UI improvements in the
classification dialog and its view widgets.
- Overhauling all accuracy assessment results, now all results based on the type
of sampling
design into three estimators: Simple/systematic estimator, Simple/systematic
post-stratified estimator
and Stratified estimator. Now the accuracy assessment result are more precise
and adjusted
to the sampling design with more statistics data such as standard deviation for
the accuracies:
overall, user and producer.
- Improvements for counting the number of pixels for each thematic value in
parallel or
lineal processing with a process dialog
- After samples are created, the file generated is prepared and selected in
respond design
and analysis tab
- Highlight the edges of the thematic pixel respectively of the current sampling
point for
the views in the response design window
- Implemented the sampling unit using a number of pixel buffer around to current
pixel for labeling,
helping the user identify the classes around the pixel that is being labelled.
- Added a new sampling method: systematic sampling using a point spacing grid
with a random offset,
but also with the option to post-stratified the sampling and the option of
neighbours aggregation.
- New dialog for easier selection the classes from categorical map to
post-stratify the simple
and systematic sampling
- Refactoring the UI, classes, functions and docs to match with the correct
names of the
accuracy assessment process described in the papers that this plugin is based
on.
- Added the option to allow AcATaMa apply an automatic classification symbology
to the layer
when the layer don't have an appropriate colors/values style
- Now the yml file save all the settings and configurations made in sampling
design tab
- Improved the nodata boxes and processes to handle the case of large and
negative nodata values
- Check if the sampling process seems stuck due to sampling conditions and ask
the user to
continue or cancel the process
- New button for reset the categorical value table, reloading the layer style,
pixel count and restoring
default values
- Checked or activated the labeling button for the label selected for the
current sample
- Adjustments and fixes in the button classification dialog
- Added a check if all button configured have a valid name in the button
classification dialog
- Set by default the classification view config by 2x1 views and readjust the
size of window
- Refactoring the save/restore configuration and state of Acatama in order to
include more configurations
- Remember the latest save/restore configuration file
- Deleted the option to save the sampling config as .ini file, use save/restore
state instead.
- Check if all button configured have a valid name and check if button name
exist (not allow duplicate
labels with the same names)
- Reviewed the user and producer accuracy value using error matrix adjusted
table for stratified and
post-stratified estimator
- Updated and adjusted the suggested filenames
- Fixes counting pixel values using parallel libraries or gdal
- Fixes some division by zero in the new way to compute the Stratified
estimator
- UI adjustments
- Cleaning unused files and libs
- Minor several fixes and improvements
- Don't forget rate the plugin :)

Previous versions:
- All commits in: https://github.com/SMByC/AcATaMa/commits