Notebooks.md

Jupyter Notebooks

In notebooks/, there are multiple Jupyter Notebooks used to test the different parts of AxoID during its development.
Note that they might not all be up-to-date and mostly contain testing bits of codes. The code in the axoid package should be the last working version.

Note: on most notebooks, they might look for model in a ./data/models/ folder not existing on the repository because it was too heavy. If it is the case, you can use the model in ./model/ instead, or re-create a ./data/models/ folder.

• AxoID_warp.ipynp: test of the entire pipeline on warped data, with automatic fine tuning
• ROI_detection_DL.ipynp: train networks, and visualize their test performance
• annotate_experiments.ipynp: use many methods to try and create annotations for full experiments (annotations are the ground truth binary detection)
• fine_tuning_test.ipynp: fine tune the network on a few manually annotated frames (annotated outside the notebook), and test results over the full experiment
• synthetic_generation.ipynp: test the functions to generate full stacks of synthetic data (see axoid.detection.synthetic)
• synthetic_tests.ipynp: test the different steps of synthetic image generation in order to see how to produce realistic looking images
• tracking_test.ipynp: test the internal model idea for the tracker. The Coherent Point Drift algorithm is also tested at the very bottom.