Documentation of WormPose.

wormpose.commands.calibrate_dataset	Calculates the image similarity on a random selection of labeled frames from a dataset.
wormpose.commands.evaluate_model	Evaluates a trained model by predicting new synthetic images and calculating the image similarity and the angle error
wormpose.commands.export_results	Exporting WormPose results to another format, for example the WCON format for a Tierpsy dataset
wormpose.commands.generate_training_data	Generates the training and evaluation data from a dataset.
wormpose.commands.postprocess_results	Post-processes WormPose results by interpolating over missing frames and smoothing
wormpose.commands.predict_dataset	Predicts videos using a trained model
wormpose.commands.train_model	Trains the neural network on the training data, supports resuming training
wormpose.commands.utils.results_saver	Save results to a H5 file
wormpose.commands.utils.time_sampling
wormpose.commands.visualize_results	Visualize the WormPose results as images showing the centerline on top of the original image
wormpose.config.default_paths	Contains constant values for default paths and folder names
wormpose.config.experiment_config
wormpose.dataset.base_dataset	This module contains the WormPose API: abstract classes to subclass in order to add a custom dataset
wormpose.dataset.features	This module deals with loading features from a dataset.
wormpose.dataset.image_processing.frame_preprocessor	Applies safely the frame preprocessing function to a frame, set the background pixels to a uniform value, deduces the region of interest
wormpose.dataset.image_processing.image_utils	Image processing example implementation for segmenting the worm in an image
wormpose.dataset.image_processing.options
wormpose.dataset.image_processing.simple_frame_preprocessing	Simple BaseFramePreprocesing implementation
wormpose.dataset.loader	The Dataset loader: instantiates the FramesDataset, FeaturesDataset, FramePreprocessing, ResultsExporter (optional).
wormpose.dataset.loaders.hdf5features	Simple FeaturesDataset implementation that reads features from a HDF5 file
wormpose.dataset.loaders.images_folder	Simple FramesDataset implementation reading image files from a folder
wormpose.dataset.loaders.resizer	Handles all about the optional resizing of the images of a Dataset
wormpose.dataset.loaders.sample_data	Dataset loader for the tutorial sample data
wormpose.dataset.loaders.tierpsy.features_dataset	Implementation of BaseFeaturesDataset to load Tierpsy tracker features
wormpose.dataset.loaders.tierpsy.frames_dataset	Implementation of BaseFramesDataset to load Tierpsy tracker frames
wormpose.dataset.loaders.tierpsy.results_exporter	Exports WormPose results (skeleton only) in WCON format
wormpose.demo.real_simple_visualizer	Visualizer for the real processed images
wormpose.demo.synthetic_simple_visualizer	Visualizer for the synthetic images
wormpose.images.real_dataset	Builds the real dataset, where images from a dataset are preprocessed and then cropped to be the same size
wormpose.images.scoring.centerline_accuracy_check	Assesses the accuracy of a centerline theta compared to a real image, by calculating the image similarity metric
wormpose.images.scoring.image_scoring	Functions that deal with the image similarity
wormpose.images.scoring.results_scoring	This module performs image scoring on the results (shuffled: two scores per frame or not) It distributes the work to several processes
wormpose.images.scoring.scoring_data_manager
wormpose.images.synthetic.synthetic_dataset	Module responsible for drawing the synthetic worm image
wormpose.images.worm_drawing	Contains the functions that deal with drawing the worm or some overlays
wormpose.machine_learning.best_models_saver	Implements a Keras callback to save the top N best models on evaluation data,
wormpose.machine_learning.eval_data_generator	Generates evaluation data: random real processed images with labels and save them to a Tfrecord file
wormpose.machine_learning.generic_file_writer	Wrapper to save the training data to different file formats
wormpose.machine_learning.loss	Definition of the loss function for the network
wormpose.machine_learning.model	Definition of the network model, a ResNet with three blocks
wormpose.machine_learning.predict_data_generator	This module assembles image data batches by preprocessing each image with a BaseFramePreprocessing object It distributes the work to several processes and provides each batch as they are ready.
wormpose.machine_learning.synthetic_data_generator	Generates synthetic data using multiprocessing
wormpose.machine_learning.tfrecord_file	Functions to read and write TFrecord files containing wormpose labeled data:
wormpose.pose.centerline	This module contains function related to the worm centerline, for example to convert skeleton (x, y) coordinates to angles, flip the head-tail orientation etc
wormpose.pose.distance_metrics	Contains function to calculate distances between worm poses, either represented as angles or as skeletons
wormpose.pose.eigenworms	Utility functions to deal with eigenworms
wormpose.pose.headtail_resolution	This module contains the logic to resolve the head-tail orientation of a predicted video time series.
wormpose.pose.postures_model	This module exposes the gaussian mixture model of worm postures from the resource file "postures_model.json.gz"
wormpose.pose.results_datatypes	This module contains data structures representing tracking results They all contain theta, scores and skeletons but we use several subclasses to specify the current step of the tracking pipeline
wormpose.version