Merge pull request #34 from LINCellularNeuroscience/vame-1.0-alpha

Vame 1.0 alpha -> Vame 1.0 RC1

VAME.yaml

@@ -24,3 +24,6 @@ dependencies:
     - pyyaml
     - opencv-python-headless
     - h5py
+    - umap-learn
+    - networkx    
+    - tqdm

examples/demo.py

@@ -1,9 +1,12 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 """
-Created on Thu Oct 31 16:50:23 2019
+Variational Animal Motion Embedding 1.0-alpha Toolbox
+© K. Luxem & P. Bauer, Department of Cellular Neuroscience
+Leibniz Institute for Neurobiology, Magdeburg, Germany
 
-@author: luxemk
+https://github.com/LINCellularNeuroscience/VAME
+Licensed under GNU General Public License v3.0
 """
 
 import vame
@@ -14,34 +17,70 @@
 videos = ['/directory/to/your/video-1','/directory/to/your/video-2','...']
 
 # Initialize your project
-# Step 1:
-config = vame.init_new_project(project=project, videos=videos, working_directory=working_directory)
+# Step 1.1:
+config = vame.init_new_project(project=project, videos=videos, working_directory=working_directory, videotype='.mp4')
 
 # After the inital creation of your project you can always access the config.yaml file 
 # via specifying the path to your project
 config = '/YOUR/WORKING/DIRECTORY/Your-VAME-Project-Apr14-2020/config.yaml'
 
+# Step 1.2:
 # Align your behavior videos egocentric and create training dataset:
-# Make sure to put them into the data folder for every video. The name of this file is the video name + -PE-seq.npy: 
-# E.g.: /Your-VAME-Project/data/video-1/video-1-PE-seq.npy
+# pose_ref_index: list of reference coordinate indices for alignment
+# Example: 0: snout, 1: forehand_left, 2: forehand_right, 3: hindleft, 4: hindright, 5: tail
+vame.egocentric_alignment(config, pose_ref_index=[0,5])
+
+# If your experiment is by design egocentrical (e.g. head-fixed experiment on treadmill etc) 
+# you can use the following to convert your .csv to a .npy array, ready to train vame on it
+vame.csv_to_numpy(config, datapath='C:\\Research\\VAME\\vame_alpha_release-Mar16-2021\\videos\\pose_estimation\\')
+
+# Step 1.3:
+# create the training set for the VAME model
 vame.create_trainset(config)
 
 # Step 2:
-# Train rnn model:
-vame.rnn_model(config, model_name='VAME', pretrained_weights=False, pretrained_model='pretrained')
+# Train VAME:
+vame.train_model(config)
 
 # Step 3:
 # Evaluate model
-vame.evaluate_model(config, model_name='VAME')
+vame.evaluate_model(config)
 
 # Step 4:
-# Quantify Behavior
-vame.behavior_segmentation(config, model_name='VAME', cluster_method='kmeans', n_cluster=[30])
+# Segment motifs/pose
+vame.pose_segmentation(config)
+
+#------------------------------------------------------------------------------
+#------------------------------------------------------------------------------
+# The following are optional choices to create motif videos, communities/hierarchies of behavior,
+# community videos
+
+# OPTIONIAL: Create motif videos to get insights about the fine grained poses
+vame.motif_videos(config, videoType='.mp4')
+
+# OPTIONAL: Create behavioural hierarchies via community detection
+vame.community(config, umap_vis=False, cut_tree=2)
+
+# OPTIONAL: Create community videos to get insights about behavior on a hierarchical scale
+vame.community_videos(config)
+
+# OPTIONAL: Down projection of latent vectors and visualization via UMAP
+vame.visualization(config, label=None) #options: label: None, "motif", "community"
+
+# OPTIONAL: Use the generative model (reconstruction decoder) to sample from 
+# the learned data distribution, reconstruct random real samples or visualize
+# the cluster center for validation
+vame.generative_model(config, mode="centers") #options: mode: "sampling", "reconstruction", "centers
+
+# OPTIONAL: Create a video of an egocentrically aligned mouse + path through 
+# the community space (similar to our gif on github) to learn more about your representation
+# and have something cool to show around ;) 
+# Note: This function is currently very slow. Once the frames are saved you can create a video
+# or gif via e.g. ImageJ or other tools
+vame.gif(config, pose_ref_index=[0,5], subtract_background=True, start=None, 
+         length=500, max_lag=30, label='community', file_format='.mp4', crop_size=(300,300))
+
+
 
-# Step 5:
-# Get behavioral transition matrix, model usage and graph
-vame.behavior_quantification(config, model_name='VAME', cluster_method='kmeans', n_cluster=30)
 
-# Get motif videos:
-vame.motif_videos(config, model_name='VAME', cluster_method="kmeans", n_cluster=[30])
 

vame/__init__.py

@@ -13,10 +13,16 @@
 
 from vame.initialize_project import init_new_project
 from vame.model import create_trainset
-from vame.model import rnn_model
+from vame.model import train_model
 from vame.model import evaluate_model
-from vame.analysis import behavior_segmentation
-from vame.analysis import behavior_quantification
+from vame.analysis import pose_segmentation
 from vame.analysis import motif_videos
+from vame.analysis import community
+from vame.analysis import community_videos
+from vame.analysis import visualization
+from vame.analysis import generative_model
+from vame.analysis import gif
 from vame.util.csv_to_npy import csv_to_numpy
+from vame.util.align_egocentrical import egocentric_alignment
 from vame.util import auxiliary
+

vame/analysis/__init__.py

@@ -11,7 +11,10 @@
 import sys
 sys.dont_write_bytecode = True
 
-from vame.analysis.segment_behavior import behavior_segmentation
-from vame.analysis.behavior_structure import behavior_quantification
-from vame.analysis.videowriter import motif_videos
+from vame.analysis.pose_segmentation import pose_segmentation
+from vame.analysis.videowriter import motif_videos, community_videos
+from vame.analysis.community_analysis import community
+from vame.analysis.umap_visualization import visualization
+from vame.analysis.generative_functions import generative_model
+from vame.analysis.gif_creator import gif