This Python package contains a graphical tool for viewing automatically classified IFCB images. It can be used to inspect and evaluate the accuracy of the classifications, as well as create new labeled image collections with the help of the classifications.
The tool is meant to be imported and used inside a Jupyter Notebook, as it relies on ipywidgets for the graphical elements. It is built with certain assumptions about how the classifications are provided (i.e., class probabilities and thresholds), since it is custom built for IFCB data processing at SYKE.
This package along with its can be installed with:
pip install .
The example Jupyter Notebook demo/viewer.ipynb demonstrates how the viewer is imported, initialized and used.
To find out what other usage options are available, see the inline documentation for JupyterViewer and its open method.
New: You can now pass in an image directory to Jupyter Viewer directly, without having to extract raw IFCB data first. This allows the tool to potentially be used with images from other instruments.