A repository containing code and data set of the paper "Automatic detection and decoding of honey bee waggle dances"
submitted the 8th of August 2017 by:

Fernando Wario
Benjamin Wild
Raul Rojas
Tim Landgraf

to PLOS ONE.

Structure:
-AttentionModule

A compiled version of the attention module, for mored details refer to S1 Text of the paper.

-FilterNetwork

Dance+Filter+Training.ipynb - A jupyter notebook with the implementation of the ConvNet used to filter WRs detected by the AM.
Results - Contains a sample of the output generated  by Dance+Filter+Training.ipynb

-OrientationModule

FourierAngle.py - Python script that implements the OM described in the paper.
Results - Contains a sample of the output generated  by FourierAngle.py

-MappingModule

Clustering.py - Python script that implements HAC and RANSAC of the MM as described in the paper.
Results - Contains a sample of the output generated  by Clustering.py
MatlabPlotter - Matlab script to plot decoded dances on a map. It works with _plot.csv files generated by Clustering.py

-GroundTruthData

GTRuns.tar - Contains a set of  200 waggle runs as extracted by the AttentionModule.

GTAverage.csv - Contains the groundtruth orientation in radians for each of the WRs in GTRuns.tar. The groundtruth value corresponds to the average
of eight manually extracted angles.