smartmove

An application for creating an Artificial Neural Network to classify body density of marine mammals from datalogger data (primarily acceleration, depth, and speed sensor data with the help of some salinity data from CTD measurements).

While currently developed primarily for the use in a specific project, the code is written as general as was possible during the project that will hopefully be easily extended and applied in future projects.

Installation

pip install git@github.com:ryanjdillon/smartmove.git

Documentation

The full documentation can be found here at Read the Docs.

Quickstart

import smartmove
smartmove.create_project('./')

# Now copy Little Leo data to `data_tag`, csv data to `data_csv`, and CTD data to `data_ctd`
# Then...

# Setup your analysis helper
a = smartmove.Analysis('./')

# Run the glide identification
a.run_glides()

# Run the ANN training
a.run_ann()

# Make the figures and tables for the paper
a.make_tables()
a.make_figures()

Contributors

• Ryan J. Dillon - All neural network code (using the theanets python package), CTD, and plotting code. Python translations of glide identification code.
• Lucia Martina Martin Lopez - Original glide identification code written in Matlab
• Kagari Aoki - Calculations of modified seal body density
• Ippei Suzuki - Propeller speed calibration calculations
• Code adapted from Stackoverflow is in the public domain, and the respective posts where it was found should be linked to in the document string of the routines in which it was used.

Acknowledgements

This development of this code was partially financed by ARCEx funded by the Research Council of Norway (Project No. 228107) and as a component of the COEXIST (Condition and energy expenditure estimates from free-ranging marine mammals project), which was financed by the Research Council of Norway (Project No. 234411/E40).

The theanets package was used for implementing the artificial neural network and is great work. Thanks Leif Johnson!

Thank you to Mark Johnson and René Swift at Sea Mammal Research Unit (St. Andrews, Scotland)the for your help and advice along the way on the processing of tag data.