* Changes and optimizations for the file IO process.
* Fixed a bug with laser pulses verification.
* Added offset parameter for laser input.
0.5.9
------------------
* Much (MUCH) faster concatenation of the data.
* Fixed a bug with the number of empty histograms that were added to the learning dataset.
0.5.8
------------------
* Robustness upgrades.
* QOL changes to GUI.
* A "power" number is needed for ``learn_histograms()`` - the percent of power given to the Qubig. It's just for saving, labeling is done with the ``label`` keyword.
* A ``foldername`` to which the data will be saved to has also been added.
0.5.7
------------------
* More robust data generation.
* Added ``scikit-learn`` to ``requirements.txt`` and ``setup.py``.
* ``CensorCorrection().learn_histograms()`` now receives the power label as its input - must be an integer.
* Return of ``CensorCorrection().learn_histograms()`` is now ``data, labels``.
* Made ``__get_bincount_deque()`` private. To be accessed using ``learn_histograms()`` only.
0.5.6
------------------
* ML classification is functional.
0.5.5
------------------
* Bug fixes for single-pixel frames
* Bug fixes for defining amount of frames manually in script.
* Censor correction shouldn't require as much memory as it did. It's still not as fast as it can be.
* Loading a configuration file will make it the "last used" file, reloading it when re-running PySight.