dappy's Introduction

3D Animal Pose Analysis in Python

Module for analysis of 3D animal pose sequences. Based on work by Berman et al. (2014) and Marshall et al. (2020).

Installation

Install the latest version of Miniconda on your machine.

The following steps will clone this repository, set up your conda environment, and install dappy.

Use environment.yml if you're on a Linux machine, and environment_osx.yml for Mac.

git clone https://github.com/joshuahwu/dappy.git
cd dappy
conda env create -n dappy -f environment.yml
conda activate dappy
conda install -c conda-forge opentsne
pip install -e .

Note that pip and setuptools must be updated to the most recent versions.

To begin gaining familiarity with the functionality of this package, download the demo dataset at this link or with the command line as follows:

cd dappy
wget -v -O ./tutorials/data/demo_mouse.h5 -L https://duke.box.com/shared/static/zprn76pl31a9u1pp6gvxbmehn7p9zmbx.h5

and run through the code in /tutorials/tutorial.ipynb.

Authors

Joshua Wu - [email protected]

dappy's People

Contributors

Stargazers

Watchers

dappy's Issues

Augmentations

Implement functions for easily creating augmentations/perturbations of features.

This includes joint ablation, speed, pitch angle, noise, and mirror augmentations.

Clustering module

Separate out watershed from embed.py and create new sub-module called cluster
Implement GMM clustering from sklearn
Test GMM clustering on wavelet transformed data

PCA Memory Efficiency

PCA implemented in features.pca() is current a memory-limiting step in the analysis pipeline. We use the fast randomized implementation in the fbpca package.

Here are some things we should implement.

Downsample frames in dataset. Calculate PC loadings on that set. Project rest of data using PC loadings.
Save PC loadings to file.
Batch updating of PC loadings w/incremental PCA
Create tests using diffsnorm to measure difference in PC scores using these approximations
Visualize eigenpostures

Recommend Projects