pose2sim/Pose2Sim/Pose2Sim.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-


'''
    ###########################################################################
    ## POSE2SIM                                                              ##
    ###########################################################################
    
    This repository offers a way to perform markerless kinematics, and gives an 
    example workflow from an Openpose input to an OpenSim result.

    It offers tools for:
    - 2D pose estimation,
    - Cameras calibration,
    - Tracking the person of interest,
    - Robust triangulation,
    - Filtration.

    It has been tested on Windows 10 but should work similarly on Linux.
    Please subscribe to this issue if you wish to be notified of the code release. 
    See https://github.com/perfanalytics/pose2sim
    
    Installation: 
    # Open Anaconda prompt. Type:
    # - conda create -n Pose2Sim python=3.7 tensorflow-gpu=1.13.1
    # - conda activate Pose2Sim
    # - conda install Pose2Sim

    Usage: 
    # First run Pose estimation and organize your directories (see Readme.md)
    from Pose2Sim import Pose2Sim
    Pose2Sim.calibration()
    Pose2Sim.personAssociation()
    Pose2Sim.triangulation()
    Pose2Sim.filtering()
    # Then run OpenSim (see Readme.md)
    
'''


## INIT
import toml
import os
import time
from copy import deepcopy
import logging, logging.handlers


## AUTHORSHIP INFORMATION
__author__ = "David Pagnon"
__copyright__ = "Copyright 2021, Pose2Sim"
__credits__ = ["David Pagnon"]
__license__ = "BSD 3-Clause License"
__version__ = "0.4"
__maintainer__ = "David Pagnon"
__email__ = "contact@david-pagnon.com"
__status__ = "Development"


## FUNCTIONS
def setup_logging(session_dir):
    '''
    Create logging file and stream handlers
    '''
    with open(os.path.join(session_dir, 'logs.txt'), 'a+') as log_f: pass
    logging.basicConfig(format='%(message)s', level=logging.INFO, 
        handlers = [logging.handlers.TimedRotatingFileHandler(os.path.join(session_dir, 'logs.txt'), when='D', interval=7), logging.StreamHandler()])

    
def recursive_update(dict_to_update, dict_with_new_values):
    '''
    Update nested dictionaries without overwriting existing keys in any level of nesting
    
    Example: 
    dict_to_update = {'key': {'key_1': 'val_1', 'key_2': 'val_2'}}
    dict_with_new_values = {'key': {'key_1': 'val_1_new'}}
    returns {'key': {'key_1': 'val_1_new', 'key_2': 'val_2'}}
    while dict_to_update.update(dict_with_new_values) would return {'key': {'key_1': 'val_1_new'}}
    '''

    for key, value in dict_with_new_values.items():
        if key in dict_to_update and isinstance(value, dict) and isinstance(dict_to_update[key], dict):
            # Recursively update nested dictionaries
            dict_to_update[key] = recursive_update(dict_to_update[key], value)
        else:
            # Update or add new key-value pairs
            dict_to_update[key] = value

    return dict_to_update


def determine_level(config_dir):
    '''
    Determine the level at which the function is called.
    Level = 1: Trial folder
    Level = 2: Participant folder
    Level = 3: Session folder
    '''

    len_paths = [len(root.split(os.sep)) for root,dirs,files in os.walk(config_dir) if 'Config.toml' in files]
    level = max(len_paths) - min(len_paths) + 1
    return level


def read_config_files(config):
    '''
    Read Session, Participant, and Trial configuration files, 
    and output a dictionary with all the parameters.
    '''

    if type(config)==dict:
        level = 3 # log_dir = os.getcwd()
        config_dicts = [config]
        if config_dicts[0].get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_PROJECT_DIRECTORY>"})')
    else:
        # if launched without an argument, config == None, else it is the path to the config directory
        config_dir = ['.' if config == None else config][0]  
        level = determine_level(config_dir)
        
        # Trial level
        if level == 1: 
            session_config_dict = toml.load(os.path.join(config_dir, '..','..','Config.toml'))
            participant_config_dict = toml.load(os.path.join(config_dir, '..','Config.toml'))
            trial_config_dict = toml.load(os.path.join(config_dir, 'Config.toml'))
                
            session_config_dict = recursive_update(session_config_dict,participant_config_dict)
            session_config_dict = recursive_update(session_config_dict,trial_config_dict)
            session_config_dict.get("project").update({"project_dir":config_dir})
            config_dicts = [session_config_dict]
        
        # Participant level
        if level == 2:
            session_config_dict = toml.load(os.path.join(config_dir, '..','Config.toml'))
            participant_config_dict = toml.load(os.path.join(config_dir, 'Config.toml'))
            config_dicts = []
            # Create config dictionaries for all trials of the participant
            for (root,dirs,files) in os.walk(config_dir):
                if 'Config.toml' in files and root != config_dir:
                    trial_config_dict = toml.load(os.path.join(root, files[0]))
                    # deep copy, otherwise session_config_dict is modified at each iteration within the config_dicts list
                    temp_dict = deepcopy(session_config_dict)
                    temp_dict = recursive_update(temp_dict,participant_config_dict)
                    temp_dict = recursive_update(temp_dict,trial_config_dict)
                    temp_dict.get("project").update({"project_dir":os.path.join(config_dir, os.path.relpath(root))})
                    if not os.path.basename(root) in temp_dict.get("project").get('exclude_from_batch'):
                        config_dicts.append(temp_dict)

        # Session level
        if level == 3:
            session_config_dict = toml.load(os.path.join(config_dir, 'Config.toml'))
            config_dicts = []
            # Create config dictionaries for all trials of all participants of the session
            for (root,dirs,files) in os.walk(config_dir):
                if 'Config.toml' in files and root != config_dir:
                    # participant
                    if determine_level(root) == 2:
                        participant_config_dict = toml.load(os.path.join(root, files[0]))
                    # trial 
                    elif determine_level(root) == 1: 
                        trial_config_dict = toml.load(os.path.join(root, files[0]))
                        # deep copy, otherwise session_config_dict is modified at each iteration within the config_dicts list
                        temp_dict = deepcopy(session_config_dict)
                        temp_dict = recursive_update(temp_dict,participant_config_dict)
                        temp_dict = recursive_update(temp_dict,trial_config_dict)
                        temp_dict.get("project").update({"project_dir":os.path.join(config_dir, os.path.relpath(root))})
                        if not os.path.relpath(root) in [os.path.relpath(p) for p in temp_dict.get("project").get('exclude_from_batch')]:
                            config_dicts.append(temp_dict)

    return level, config_dicts


def calibration(config=None):
    '''
    Cameras calibration from checkerboards or from qualisys files.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''

    from Pose2Sim.calibration import calibrate_cams_all

    level, config_dicts = read_config_files(config)
    config_dict = config_dicts[0]
    session_dir = os.path.realpath([os.getcwd() if level==3 else os.path.join(os.getcwd(), '..') if level==2 else os.path.join(os.getcwd(), '..', '..')][0])
    config_dict.get("project").update({"project_dir":session_dir})

    # Set up logging
    setup_logging(session_dir)  
    
    # Run calibration
    calib_dir = [os.path.join(session_dir, c) for c in os.listdir(session_dir) if ('Calib' or 'calib') in c][0]
    logging.info("\n\n---------------------------------------------------------------------")
    logging.info("Camera calibration")
    logging.info("---------------------------------------------------------------------")
    logging.info(f"\nCalibration directory: {calib_dir}")
    start = time.time()
    
    calibrate_cams_all(config_dict)
    
    end = time.time()
    logging.info(f'Calibration took {end-start:.2f} s.')


def poseEstimation(config=None):
    '''
    Estimate pose using BlazePose, OpenPose, AlphaPose, or DeepLabCut.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''
    
    raise NotImplementedError('This has not been integrated yet. \nPlease read README.md for further explanation')
    
    # TODO
    # Determine the level at which the function is called (session:3, participant:2, trial:1)
    level, config_dicts = read_config_files(config)

    if type(config)==dict:
        config_dict = config_dicts[0]
        if config_dict.get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_TRIAL_DIRECTORY>"})')

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)    

    # Batch process all trials
    for config_dict in config_dicts:
        start = time.time()
        project_dir = os.path.realpath(config_dict.get('project').get('project_dir'))
        seq_name = os.path.basename(project_dir)
        frame_range = config_dict.get('project').get('frame_range')
        frames = ["all frames" if frame_range == [] else f"frames {frame_range[0]} to {frame_range[1]}"][0]

        logging.info("\n\n---------------------------------------------------------------------")
        logging.info("Camera synchronization")
        logging.info("---------------------------------------------------------------------")
        logging.info(f"\nProject directory: {project_dir}")
    
        pose_estimation_all(config_dict)
        
        end = time.time()
        logging.info(f'Pose estimation took {end-start:.2f} s.')
    

def synchronization(config=None):
    '''
    Synchronize cameras if needed.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''   

    raise NotImplementedError('This has not been integrated yet. \nPlease read README.md for further explanation')
    
    #TODO
    # Determine the level at which the function is called (session:3, participant:2, trial:1)
    level, config_dicts = read_config_files(config)

    if type(config)==dict:
        config_dict = config_dicts[0]
        if config_dict.get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_TRIAL_DIRECTORY>"})')

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)    

    # Batch process all trials
    for config_dict in config_dicts:
        start = time.time()
        project_dir = os.path.realpath(config_dict.get('project').get('project_dir'))
        seq_name = os.path.basename(project_dir)
        frame_range = config_dict.get('project').get('frame_range')
        frames = ["all frames" if frame_range == [] else f"frames {frame_range[0]} to {frame_range[1]}"][0]

        logging.info("\n\n---------------------------------------------------------------------")
        logging.info("Camera synchronization")
        logging.info("---------------------------------------------------------------------")
        logging.info(f"\nProject directory: {project_dir}")
        
        synchronize_cams_all(config_dict)
    
        end = time.time()
        logging.info(f'Synchronization took {end-start:.2f} s.')    
    
    
def personAssociation(config=None):
    '''
    Tracking one or several persons of interest.
    Needs a calibration file.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''
    
    from Pose2Sim.personAssociation import track_2d_all

    # Determine the level at which the function is called (session:3, participant:2, trial:1)
    level, config_dicts = read_config_files(config)

    if type(config)==dict:
        config_dict = config_dicts[0]
        if config_dict.get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_TRIAL_DIRECTORY>"})')

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)    

    # Batch process all trials
    for config_dict in config_dicts:
        start = time.time()
        project_dir = os.path.realpath(config_dict.get('project').get('project_dir'))
        seq_name = os.path.basename(project_dir)
        frame_range = config_dict.get('project').get('frame_range')
        frames = ["all frames" if frame_range == [] else f"frames {frame_range[0]} to {frame_range[1]}"][0]

        logging.info("\n\n---------------------------------------------------------------------")
        logging.info(f"Associating persons for {seq_name}, for {frames}.")
        logging.info("---------------------------------------------------------------------")
        logging.info(f"\nProject directory: {project_dir}")
    
        track_2d_all(config_dict)
    
        end = time.time()
        logging.info(f'Associating persons took {end-start:.2f} s.')
    
    
def triangulation(config=None):
    '''
    Robust triangulation of 2D points coordinates.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''

    from Pose2Sim.triangulation import triangulate_all

    # Determine the level at which the function is called (session:3, participant:2, trial:1)
    level, config_dicts = read_config_files(config)

    if type(config)==dict:
        config_dict = config_dicts[0]
        if config_dict.get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_TRIAL_DIRECTORY>"})')

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)  

    # Batch process all trials
    for config_dict in config_dicts:
        start = time.time()
        project_dir = os.path.realpath(config_dict.get('project').get('project_dir'))
        seq_name = os.path.basename(project_dir)
        frame_range = config_dict.get('project').get('frame_range')
        frames = ["all frames" if frame_range == [] else f"frames {frame_range[0]} to {frame_range[1]}"][0]

        logging.info("\n\n---------------------------------------------------------------------")
        logging.info(f"Triangulation of 2D points for {seq_name}, for {frames}.")
        logging.info("---------------------------------------------------------------------")
        logging.info(f"\nProject directory: {project_dir}")
    
        triangulate_all(config_dict)
    
    end = time.time()
    logging.info(f'Triangulation took {end-start:.2f} s.')
    
    
def filtering(config=None):
    '''
    Filter trc 3D coordinates.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''

    from Pose2Sim.filtering import filter_all

    # Determine the level at which the function is called (session:3, participant:2, trial:1)
    level, config_dicts = read_config_files(config)

    if type(config)==dict:
        config_dict = config_dicts[0]
        if config_dict.get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_TRIAL_DIRECTORY>"})')

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)    

    # Batch process all trials
    for config_dict in config_dicts:
        project_dir = os.path.realpath(config_dict.get('project').get('project_dir'))
        seq_name = os.path.basename(project_dir)
        frame_range = config_dict.get('project').get('frame_range')
        frames = ["all frames" if frame_range == [] else f"frames {frame_range[0]} to {frame_range[1]}"][0]
    
        logging.info("\n\n---------------------------------------------------------------------")
        logging.info(f"Filtering 3D coordinates for {seq_name}, for {frames}.")
        logging.info("---------------------------------------------------------------------")
        logging.info(f"\nProject directory: {project_dir}")
    
        filter_all(config_dict)


def opensimProcessing(config=None):
    '''
    Uses OpenSim to run scaling based on a static trc pose
    and inverse kinematics based on a trc motion file.
    
    config can be a dictionary,
    or a the directory path of a trial, participant, or session,
    or the function can be called without an argument, in which case it the config directory is the current one.
    '''
    
    raise NotImplementedError('This has not been integrated yet. \nPlease read README.md for further explanation')
    
    # TODO
    from Pose2Sim.opensimProcessing import opensim_processing_all
    
    # Determine the level at which the function is called (session:3, participant:2, trial:1)
    level, config_dicts = read_config_files(config)

    if type(config)==dict:
        config_dict = config_dicts[0]
        if config_dict.get('project').get('project_dir') == None:
            raise ValueError('Please specify the project directory in config_dict:\n \
                             config_dict.get("project").update({"project_dir":"<YOUR_TRIAL_DIRECTORY>"})')

    # Set up logging
    session_dir = os.path.realpath(os.path.join(config_dicts[0].get('project').get('project_dir'), '..', '..'))
    setup_logging(session_dir)    

    # Batch process all trials
    for config_dict in config_dicts:
        start = time.time()
        project_dir = os.path.realpath(config_dict.get('project').get('project_dir'))
        seq_name = os.path.basename(project_dir)
        frame_range = config_dict.get('project').get('frame_range')
        frames = ["all frames" if frame_range == [] else f"frames {frame_range[0]} to {frame_range[1]}"][0]

        logging.info("\n\n---------------------------------------------------------------------")
        # if static_file in project_dir: 
        #     logging.info(f"Scaling model with <STATIC TRC FILE>.")
        # else:
        #     logging.info(f"Running inverse kinematics <MOTION TRC FILE>.")
        logging.info("---------------------------------------------------------------------")
        logging.info(f"\nOpenSim output directory: {project_dir}")
   
        opensim_processing_all(config_dict)
    
        end = time.time()
        logging.info(f'Model scaling took {end-start:.2f} s.')