v3.0

Allows multi-class image classification, cleaned and sorted.

v2.1

Version used for rhododendron binary classification.

2.0

Major back end and API upgrades for compatibility with TF 2.5 (with Python 3.9). More in line with standard design patterns, executes much more quickly. (~1/3 the previous runtime.)

Cross-fold validation has not yet been implemented in this version. All other functionality is working.

v1.1

Minor updates needed for Tensorflow 2.x, but nothing that changed the usage of this repo, or any higher-level coding functionality:

• keras.models.save_model() usage changed to function
• some changes around the random seed/compatibility mode

All utilities/archives files were simply changed to import v1 compatibility mode as tf.

v1.0

Code used for classifying cryptic species of Frullania (F. rostrata and an undescribed species).