bobby/open_dbm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactoring process_data.py

Browse Source
This commit is contained in:
jordi.hasianta
2022-09-16 00:42:49 +07:00
parent e41336309c
commit 36e70e44c9
1 changed files with 166 additions and 89 deletions
Download Patch File Download Diff File Expand all files Collapse all files

255
process_data.py
View File

@@ -4,27 +4,29 @@ project_name: DBM
created: 2020-20-07 created: 2020-20-07
""" """
from opendbm.dbm_lib.config import config_reader, config_derive_feature, config_raw_feature
from opendbm.dbm_lib.controller import process_feature as pf
from opendbm.dbm_lib import open_face_process as of
from opendbm.dbm_lib.dbm_features.derived_features import derive as der
import pandas as pd
import os
import argparse import argparse
import logging
import glob import glob
import time import logging
import os
import subprocess import subprocess
import time
from os.path import splitext from os.path import splitext
logging.basicConfig(level=logging.INFO) import pandas as pd
logger=logging.getLogger()
from opendbm.dbm_lib import config_derive_feature, config_raw_feature, config_reader
from opendbm.dbm_lib.controller import process_feature as pf
from opendbm.dbm_lib.dbm_features.derived_features import derive as der
from opendbm.dbm_lib.dbm_features.raw_features.video import open_face_process as of
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger()
OPENFACE_PATH_VIDEO = "opendbm/pkg/open_dbm/OpenFace/build/bin/FaceLandmarkVid"
OPENFACE_PATH = "opendbm/pkg/open_dbm/OpenFace/build/bin/FeatureExtraction"
DEEP_SPEECH = "opendbm/pkg/DeepSpeech"
DLIB_SHAPE_MODEL = "opendbm/pkg/shape_detector/shape_predictor_68_face_landmarks.dat"
OPENFACE_PATH_VIDEO = 'opendbm/pkg/open_dbm/OpenFace/build/bin/FaceLandmarkVid'
OPENFACE_PATH = 'opendbm/pkg/open_dbm/OpenFace/build/bin/FeatureExtraction'
DEEP_SPEECH = 'pkg/DeepSpeech'
DLIB_SHAPE_MODEL = 'pkg/shape_detector/shape_predictor_68_face_landmarks.dat'
def common_video(video_file, args, r_config): def common_video(video_file, args, r_config):
""" """
@@ -34,18 +36,41 @@ def common_video(video_file, args, r_config):
args: user supplied arguments args: user supplied arguments
r_config: raw feature config object r_config: raw feature config object
""" """
out_path = os.path.join(args.output_path, 'raw_variables') out_path = os.path.join(args.output_path, "raw_variables")
pf.audio_to_wav(video_file) pf.audio_to_wav(video_file)
of.process_open_face(video_file, os.path.dirname(video_file), out_path, OPENFACE_PATH, args.dbm_group,video_tracking=False) of.process_open_face(
video_file,
os.path.dirname(video_file),
out_path,
OPENFACE_PATH,
args.dbm_group,
video_tracking=False,
)
pf.process_facial(video_file, out_path, args.dbm_group, r_config) pf.process_facial(video_file, out_path, args.dbm_group, r_config)
pf.process_acoustic(video_file, out_path, args.dbm_group, r_config) pf.process_acoustic(video_file, out_path, args.dbm_group, r_config)
pf.process_nlp(video_file, out_path, args.dbm_group, args.tr, r_config, DEEP_SPEECH) pf.process_nlp(video_file, out_path, args.dbm_group, args.tr, r_config, DEEP_SPEECH)
if args.dbm_group == None or len(args.dbm_group)>0 and 'movement' in args.dbm_group:
of.process_open_face(video_file, os.path.dirname(video_file), out_path, OPENFACE_PATH_VIDEO, args.dbm_group, video_tracking=True) if (
pf.process_movement(video_file, out_path, args.dbm_group, r_config, DLIB_SHAPE_MODEL) args.dbm_group is None
or len(args.dbm_group) > 0
and "movement" in args.dbm_group
):
of.process_open_face(
video_file,
os.path.dirname(video_file),
out_path,
OPENFACE_PATH_VIDEO,
args.dbm_group,
video_tracking=True,
)
pf.process_movement(
video_file, out_path, args.dbm_group, r_config, DLIB_SHAPE_MODEL
)
pf.remove_file(video_file) pf.remove_file(video_file)
def process_raw_video_file(args, s_config, r_config): def process_raw_video_file(args, s_config, r_config):
""" """
Processing video file Processing video file
@@ -55,20 +80,22 @@ def process_raw_video_file(args, s_config, r_config):
r_config: raw feature config object r_config: raw feature config object
""" """
try: try:
if args.output_path != None: if args.output_path is not None:
video_file = glob.glob(args.input_path) video_file = glob.glob(args.input_path)
if len(video_file)>0: if len(video_file) > 0:
logger.info('Calculating raw variables...') logger.info("Calculating raw variables...")
common_video(video_file[0], args, r_config) common_video(video_file[0], args, r_config)
else: else:
logger.info('Enter correct video(*.mp4) file path.') logger.info("Enter correct video(*.mp4) file path.")
except Exception as e: except Exception as e:
logger.error('Failed to process mp4 file.') e
logger.error("Failed to process mp4 file.", str(e))
pf.remove_file(video_file[0]) pf.remove_file(video_file[0])
def process_raw_audio_file(args, s_config, r_config): def process_raw_audio_file(args, s_config, r_config):
""" """
Processing audio file Processing audio file
@@ -78,19 +105,28 @@ def process_raw_audio_file(args, s_config, r_config):
r_config: raw feature config object r_config: raw feature config object
""" """
try: try:
if args.output_path != None: if args.output_path is not None:
audio_file = glob.glob(args.input_path) audio_file = glob.glob(args.input_path)
if len(audio_file)>0: if len(audio_file) > 0:
logger.info('Calculating raw variables...') logger.info("Calculating raw variables...")
out_path = os.path.join(args.output_path, 'raw_variables') out_path = os.path.join(args.output_path, "raw_variables")
pf.process_acoustic(audio_file[0], out_path, args.dbm_group, r_config) pf.process_acoustic(audio_file[0], out_path, args.dbm_group, r_config)
pf.process_nlp(audio_file[0], out_path, args.dbm_group, args.tr, r_config, DEEP_SPEECH) pf.process_nlp(
audio_file[0],
out_path,
args.dbm_group,
args.tr,
r_config,
DEEP_SPEECH,
)
else: else:
logger.info('Enter correct audio(*.wav) file path.') logger.info("Enter correct audio(*.wav) file path.")
except Exception as e: except Exception as e:
logger.error('Failed to process wav file.') # e
logger.error("Failed to process wav file.", str(e))
def process_raw_video_dir(args, s_config, r_config): def process_raw_video_dir(args, s_config, r_config):
""" """
@@ -100,27 +136,35 @@ def process_raw_video_dir(args, s_config, r_config):
s_config: service config object s_config: service config object
r_config: raw feature config object r_config: raw feature config object
""" """
if args.output_path != None: if args.output_path is not None:
vid_loc = glob.glob(args.input_path + '/*.mp4') + glob.glob(args.input_path + '/*.mov') + glob.glob(args.input_path + '/*.MOV') vid_loc = (
glob.glob(args.input_path + "/*.mp4")
+ glob.glob(args.input_path + "/*.mov")
+ glob.glob(args.input_path + "/*.MOV")
)
if len(vid_loc) == 0: if len(vid_loc) == 0:
logger.info('Directory does not have any MP4 files.') logger.info("Directory does not have any MP4 files.")
return return
logger.info('Calculating raw variables...') logger.info("Calculating raw variables...")
for vid_file in vid_loc: for vid_file in vid_loc:
try: try:
fname, file_ext = os.path.splitext(vid_file) fname, file_ext = os.path.splitext(vid_file)
if file_ext.lower() == '.mov': if file_ext.lower() == ".mov":
convert_file(vid_file) convert_file(vid_file)
common_video(fname+'.mp4', args, r_config) common_video(fname + ".mp4", args, r_config)
remove_convert(vid_file, '.mp4') #removing files(ffmpeg converted ) after processing remove_convert(
vid_file, ".mp4"
) # removing files(ffmpeg converted ) after processing
except Exception as e: except Exception as e:
logger.error('Failed to process mp4 file.') e
logger.error("Failed to process mp4 file.", str(e))
pf.remove_file(vid_file) pf.remove_file(vid_file)
def process_raw_audio_dir(args, s_config, r_config): def process_raw_audio_dir(args, s_config, r_config):
""" """
Processing audio file Processing audio file
@@ -129,27 +173,42 @@ def process_raw_audio_dir(args, s_config, r_config):
s_config: service config object s_config: service config object
r_config: raw feature config object r_config: raw feature config object
""" """
if args.output_path != None: if args.output_path is not None:
audio_loc = glob.glob(args.input_path + '/*.wav') + glob.glob(args.input_path + '/*.mp3') + glob.glob(args.input_path + '/*.MP3') audio_loc = (
glob.glob(args.input_path + "/*.wav")
+ glob.glob(args.input_path + "/*.mp3")
+ glob.glob(args.input_path + "/*.MP3")
)
if len(audio_loc) == 0: if len(audio_loc) == 0:
logger.info('Directory does not have any WAV files.') logger.info("Directory does not have any WAV files.")
return return
logger.info('Calculating raw variables...') logger.info("Calculating raw variables...")
for audio in audio_loc: for audio in audio_loc:
try: try:
fname, file_ext = os.path.splitext(audio) fname, file_ext = os.path.splitext(audio)
if file_ext.lower() == '.mp3': if file_ext.lower() == ".mp3":
convert_file(audio) convert_file(audio)
out_path = os.path.join(args.output_path, 'raw_variables') out_path = os.path.join(args.output_path, "raw_variables")
pf.process_acoustic(fname+'.wav', out_path, args.dbm_group, r_config) pf.process_acoustic(fname + ".wav", out_path, args.dbm_group, r_config)
pf.process_nlp(fname +'.wav', out_path, args.dbm_group, args.tr, r_config, DEEP_SPEECH) pf.process_nlp(
fname + ".wav",
remove_convert(audio, '.wav') #removing files(ffmpeg converted) after processing out_path,
args.dbm_group,
args.tr,
r_config,
DEEP_SPEECH,
)
remove_convert(
audio, ".wav"
) # removing files(ffmpeg converted) after processing
except Exception as e: except Exception as e:
logger.error('Failed to process wav file.') e
logger.error("Failed to process wav file.")
def convert_file(input_filepath): def convert_file(input_filepath):
""" """
@@ -158,52 +217,70 @@ def convert_file(input_filepath):
_, file_ext = os.path.splitext(os.path.basename(input_filepath)) _, file_ext = os.path.splitext(os.path.basename(input_filepath))
fname, _ = splitext(input_filepath) fname, _ = splitext(input_filepath)
call = [] call = []
if file_ext.lower() == '.mp3':
output_filepath = fname + '.wav'
logger.info('Converting audio from {} to wav'.format(input_filepath))
call = ['ffmpeg', '-i', input_filepath, output_filepath]
if file_ext.lower() == '.mov': if file_ext.lower() == ".mp3":
output_filepath = fname + '.mp4' output_filepath = fname + ".wav"
logger.info('Converting video from {} to mp4'.format(input_filepath)) logger.info("Converting audio from {} to wav".format(input_filepath))
call = ['ffmpeg', '-i', input_filepath, '-vcodec', 'h264','-acodec','aac', '-strict', '-2', output_filepath] call = ["ffmpeg", "-i", input_filepath, output_filepath]
if len(call)>0: if file_ext.lower() == ".mov":
output_filepath = fname + ".mp4"
logger.info("Converting video from {} to mp4".format(input_filepath))
call = [
"ffmpeg",
"-i",
input_filepath,
"-vcodec",
"h264",
"-acodec",
"aac",
"-strict",
"-2",
output_filepath,
]
if len(call) > 0:
subprocess.check_output(call) subprocess.check_output(call)
def remove_convert(input_filepath, file_ext): def remove_convert(input_filepath, file_ext):
""" """
removing converted files after processing removing converted files after processing
""" """
expected_ext = ['.mp3', '.mov'] expected_ext = [".mp3", ".mov"]
input_loc, inp_ext = os.path.splitext(input_filepath) input_loc, inp_ext = os.path.splitext(input_filepath)
if inp_ext.lower() in expected_ext: if inp_ext.lower() in expected_ext:
pf.remove_file(input_loc + file_ext, file_ext) pf.remove_file(input_loc + file_ext, file_ext)
def process_derive(args, r_config, d_config, input_type): def process_derive(args, r_config, d_config, input_type):
""" """
Processing dbm derived variables Processing dbm derived variables
""" """
if input_type == 'file': if input_type == "file":
input_file = glob.glob(args.input_path) input_file = glob.glob(args.input_path)
else: else:
input_file = glob.glob(args.input_path + '/*') input_file = glob.glob(args.input_path + "/*")
out_raw_path = os.path.join(args.output_path, 'raw_variables') out_raw_path = os.path.join(args.output_path, "raw_variables")
out_derive_path = os.path.join(args.output_path, 'derived_variables') out_derive_path = os.path.join(args.output_path, "derived_variables")
logger.info('Calculating derived variables...') logger.info("Calculating derived variables...")
feature_df = der.run_derive(input_file, out_raw_path, out_derive_path, r_config, d_config) der.run_derive(input_file, out_raw_path, out_derive_path, r_config, d_config)
if __name__=="__main__":
if __name__ == "__main__":
start_time = time.time() start_time = time.time()
parser = argparse.ArgumentParser(description="Process video/audio......") parser = argparse.ArgumentParser(description="Process video/audio......")
parser.add_argument("--input_path", help="path to the input files", required=True) parser.add_argument("--input_path", help="path to the input files", required=True)
parser.add_argument("--output_path", help="path to the raw and derived variable output", required=True) parser.add_argument(
parser.add_argument("--dbm_group", help="list of feature groups", nargs='+') "--output_path",
help="path to the raw and derived variable output",
required=True,
)
parser.add_argument("--dbm_group", help="list of feature groups", nargs="+")
parser.add_argument("--tr", help="speech transcription toogle") parser.add_argument("--tr", help="speech transcription toogle")
args = parser.parse_args() args = parser.parse_args()
@@ -214,28 +291,28 @@ if __name__=="__main__":
_, file_ext = os.path.splitext(os.path.basename(args.input_path)) _, file_ext = os.path.splitext(os.path.basename(args.input_path))
if file_ext: if file_ext:
input_type = 'file' input_type = "file"
if file_ext.lower() in ['.mp4','.mov']: if file_ext.lower() in [".mp4", ".mov"]:
if file_ext.lower() == '.mov': if file_ext.lower() == ".mov":
convert_file(args.input_path) convert_file(args.input_path)
process_raw_video_file(args, s_config, r_config) process_raw_video_file(args, s_config, r_config)
remove_convert(args.input_path, '.mp4') remove_convert(args.input_path, ".mp4")
elif file_ext.lower() in ['.wav','.mp3']: elif file_ext.lower() in [".wav", ".mp3"]:
if file_ext.lower() == '.mp3': if file_ext.lower() == ".mp3":
convert_file(args.input_path) convert_file(args.input_path)
process_raw_audio_file(args, s_config, r_config) process_raw_audio_file(args, s_config, r_config)
remove_convert(args.input_path, '.wav') remove_convert(args.input_path, ".wav")
else: else:
logger.error('No WAV/MP3 or MOV/MP4 files detected in input path') logger.error("No WAV/MP3 or MOV/MP4 files detected in input path")
else: else:
input_type = 'dir' input_type = "dir"
process_raw_video_dir(args, s_config, r_config) process_raw_video_dir(args, s_config, r_config)
process_raw_audio_dir(args, s_config, r_config) process_raw_audio_dir(args, s_config, r_config)
process_derive(args, r_config, d_config, input_type) process_derive(args, r_config, d_config, input_type)
exec_time = time.time() - start_time exec_time = time.time() - start_time
logger.info('Done! Processing time: {} seconds'.format(exec_time)) logger.info("Done! Processing time: {} seconds".format(exec_time))