def __init__(self, scenario_name: str):
self.scenario_name = scenario_name
fm = File_Manager()
self.scenario_cfgs = fm.read_scenario_config(self.scenario_name)
self.current_epoch = self.get_cfgs('current_epoch', default=0)
self.optimizer_type = self.get_cfgs('optimizer_type')
self.scenes = OrderedDict({})
for scene_cfgs in self.get_cfgs('scenes'):
scene_name = scene_cfgs['name']
task_defaults = {}
if 'task_defaults' in scene_cfgs:
task_defaults = scene_cfgs['task_defaults']
scene_cfgs['tasks'] = {
t: fm.read_task_config(task_name=t,
scenario=self,
scene_defaults=task_defaults)
for t in scene_cfgs['tasks']
}
self.scenes[scene_name] = scene_cfgs
self.scenario_lengths = [{
'name': key,
'len': len(s)
} for (key, s) in self.scenes.items()]
self.collect_dictionaries()
self.__current_scene = None
def end_epoch(self, summary: dict, scene_name: str):
summary['dataset_name'] = self.dataset_name
summary['experiment_name'] = self.experiment_name
summary['modalities'] = defaultdict(dict)
raw_csv_data = []
for _, modality in self.modalities.items():
raw_csv_data.extend(modality.get_runtime_values())
modality.report_epoch_summary(summary)
if modality.is_explicit_modality() and modality.is_csv():
if isinstance(modality.content, pd.Series):
raw_csv_data.append(modality.content)
elif isinstance(modality.content, pd.DataFrame):
[
raw_csv_data.append(modality.content[c])
for c in modality.content
]
else:
raise ValueError(
f'The content type of {modality.get_name()} is not implemented'
)
# We want the column estimates to be close to eachother in the final output
raw_csv_data.sort(key=lambda v: v.name if hasattr(v, 'name') else -1)
File_Manager().write_csv_annotation(
annotations=pd.concat(raw_csv_data, axis=1),
dataset_name=self.dataset_name,
experiment_file_name=
f'{scene_name}_{self.get_cfgs("annotations_path")}',
)
return summary
def init_dictionary(self):
if self.dictionary is None:
raise Exception(
f'No dictionary has been initated for {self.get_name()}')
File_Manager().write_dictionary2logdir(dictionary=self.dictionary,
modality_name=self.get_name())
def setup_annotations(self):
rel_path = self.get_cfgs('annotations_path')
fm = File_Manager()
self.annotations = fm.read_csv_annotations(
dataset_name=self.dataset_name,
annotations_rel_path=rel_path,
multi_view_per_sample=self.multi_view_per_sample,
)
if self.annotations is None:
annotations_url = self.get_cfgs('annotations_url')
available_csvs_str = '\', \''.join(
fm.get_available_csvs(self.dataset_name))
Console_UI().inform_user(
'"%s" does not exist among the available datasets: \'%s\'.\nDownloading from:\n %s'
% (rel_path, available_csvs_str, annotations_url))
fm.download_zip_file(
url=annotations_url,
dataset_name=self.get_dataset_name(),
)
self.annotations = fm.read_csv_annotations(
dataset_name=self.dataset_name,
annotations_rel_path=rel_path,
multi_view_per_sample=self.multi_view_per_sample,
)
if self.get_cfgs('test_run'):
self.annotations = self.annotations[
self.annotations.index.get_level_values(0) < 100]
def __init__(self, test_mode=False, test_init_values={}):
if test_mode:
self.cfgs = test_init_values
Console_UI(self.get('log_level', 'warning'), globalConfigs=self)
self.sub_log_path = self.get('sub_log_path', 'sub_log_not_set')
File_Manager(annotations_root=self.get('annotation_root'),
log_folder=self.get('log_folder'),
scenario_log_root=self.get('scenario_log_root'),
resume_prefix=self.get('resume'),
resume_scene=self.get('resume_scene'),
tmp_root=self.get('tmp_root'),
model_zoo_root=self.get('model_zoo_root'),
global_cfgs=self)
return
args = self.parse_argument()
self.cfgs = args.__dict__
Console_UI(self.get('log_level', 'info'), globalConfigs=self)
self.read_environment_variables()
self.start_time = datetime.now(
pytz.timezone('Europe/Stockholm')).strftime('%Y%m%d/%H.%M')
self.sub_log_path = os.path.join(self.get('scenario'), self.start_time)
if self.get('resume') is not None:
self.prep_resume()
fm = File_Manager(scenario_log_root=self.scenario_log_root,
log_folder=self.log_folder,
annotations_root=self.get('annotation_root'),
resume_prefix=self.get('resume'),
resume_scene=self.get('resume_scene'),
tmp_root=self.get('tmp_root'),
model_zoo_root=self.get('model_zoo_root'),
global_cfgs=self)
setup_data = {'call': ' '.join(sys.argv), 'setup': self.cfgs}
fm.log_setup(data=setup_data, name='base_setup.yaml')
self.__forward_noise = 0
def load(self):
neural_net_name = self.get_name()
state_dict = File_Manager().load_pytorch_neural_net(
neural_net_name=neural_net_name)
if state_dict is not None:
try:
# As the save is done at the layers level: neural_net.layers.state_dict()
# we need to load it from the layers
self.layers.load_state_dict(state_dict)
except RuntimeError as e:
raise RuntimeError(
f'Failed to load dictionary for {neural_net_name} \nError message: {e}'
)
def init_dictionary(self):
if self.get_cfgs('skip_dictionary_save', default=False):
return
column_dictionary = pd.DataFrame({
'columns': [c for c in self.csv_columns],
'labels': [self.column_map[c] for c in self.csv_columns],
'index':
range(len(self.csv_columns))
})
File_Manager().write_dictionary2logdir(dictionary=column_dictionary,
modality_name=self.get_name())
def append_suggested_dictionary(self,
dataset_name,
modality_name,
FMSingleton=None):
assert isinstance(
modality_name,
str), f'Modality name is not a string "{modality_name}"'
assert isinstance(
dataset_name,
str), f'Dataset name is not a string "{modality_name}"'
modality_name = modality_name.lower()
if FMSingleton is None:
FMSingleton = File_Manager()
suggested_dictionary = FMSingleton.read_dictionary(
dataset_name=dataset_name,
modality_name=modality_name,
)
if suggested_dictionary is not None:
self.__suggested_dictionaries[modality_name].append(
suggested_dictionary)
return self
def init_dictionary(self):
if self.dictionary is None:
fm = File_Manager()
self.dictionary = fm.read_dictionary(
dataset_name=self.dataset_name, modality_name=self.get_name())
if self.dictionary is None:
self.dictionary = self.make_dictionary() # no dictionary
fm.write_dictionary(dictionary=self.dictionary,
dataset_name=self.dataset_name,
modality_name=self.get_name())
else:
fm.write_dictionary2logdir(dictionary=self.dictionary,
modality_name=self.get_name())
def run(config_manager, logger):
logger.msg('loading configuration file...')
config = config_manager.get_config()
logger.msg('ensuring that file directories exist...')
file_manager = File_Manager(config['speech'], config['text'])
if (not file_manager.directories_exist()):
logger.err(
'Please check the directories entered into the config file: ' +
config_file_name)
return
logger.msg('waiting for files to transcribe...')
transcriber = Transcriber(config['username'], config['password'],
file_manager) # Create transcription object
while (True): # Loop for eternity
for file in file_manager.speech_files_without_text_files(
): # Loop through files that need to be transcribed
logger.msg('Transcribing: ' + file + '...')
transcriber.transcribe(file) # Transcribe the current file
logger.msg(file + ' transcribed')
logger.msg('waiting for files to transcribe...')
sleep(10) # Wait for 10 seconds
def get_dataset_cfgs(self, dataset_name):
dataset_cfgs = File_Manager().read_dataset_config(dataset_name)
# For the sake of simplicity, when modalities are identical during
# train and tests, we can just write "modalities": "same_as_X" in
# the config file.(in this example, X is "train")
# This piece of code searches for the modalities like this and
# replace them with the "X" modalities
for _, experiment_cfgs in dataset_cfgs['experiments'].items():
if (isinstance(experiment_cfgs['modalities'], str)
and experiment_cfgs['modalities'].startswith('same_as_')):
other_experiment = experiment_cfgs['modalities'][len('same_as_'):]
experiments = dataset_cfgs['experiments']
if other_experiment in experiments:
experiment_cfgs['modalities'] = experiments[other_experiment]['modalities']
else:
raise KeyError('Could not find the modality \'%s\' among the modalities: \'%s\'' %
(other_experiment, '\', \''.join(experiments.keys())))
return dataset_cfgs
def get_dataset(
self,
dataset_name,
batch_size_multiplier: float,
):
fixed_name = dataset_name.lower()
if fixed_name not in self.datasets:
predefined_datasets = File_Manager().get_dataset_definitions()
if (fixed_name in predefined_datasets):
from .csv_dataset import CSV_Dataset as Dataset
else:
raise Exception('The dataset \'%s\' is not among the predefined sets: \'%s\'' %
(dataset_name, '\', \''.join(predefined_datasets)))
self.datasets[fixed_name] = Dataset(
dataset_name=fixed_name,
batch_size_multiplier=batch_size_multiplier,
)
else:
self.datasets[fixed_name].set_batch_size_multiplier(batch_size_multiplier)
return self.datasets[fixed_name]
def save(self, scene_name='last'):
File_Manager().save_pytorch_neural_net(self.get_name(), self,
scene_name)
def collect_dictionaries(self):
"""
Check all the Datasets for common items, e.g. body part and then create
a general dictionary for all of them.
"""
datasets = []
for scene in self.scenario_cfgs['scenes']:
for task in scene['tasks'].values():
if task['dataset_name'] not in datasets:
datasets.append(task['dataset_name'])
configs = {}
for dataset_name in datasets:
configs[dataset_name] = File_Manager().read_dataset_config(
dataset_name)
modalities_with_dictionaries = [
'one_vs_rest',
'bipolar',
'multi_bipolar',
] # TODO: add 'hierarchical_label' but this has some fancy logic :-S
dictionary_candidates = []
for dataset_name in datasets:
config = configs[dataset_name]
try:
for experiment in config['experiments'].values():
if isinstance(experiment['modalities'], dict):
[
dictionary_candidates.append(name)
for name, cfg in experiment['modalities'].items()
if
cfg['type'].lower() in modalities_with_dictionaries
and name not in dictionary_candidates
]
except Exception as e:
raise Exception(
f'Failed to get dictionary for {dataset_name}: {e}')
# Store all the different values available for this modality into the dictionary singleton that
# keeps track of the unique values
dg = Dictionary_Generator()
for modality_name in dictionary_candidates:
for dataset_name in datasets:
dg.append_suggested_dictionary(dataset_name=dataset_name,
modality_name=modality_name)
config = configs[dataset_name]
for experiment in config['experiments'].values():
annotations = File_Manager().read_csv_annotations(
dataset_name,
annotations_rel_path=experiment['annotations_path'],
# Multi-view argument should be irrelevant for this
)
if annotations is None:
raise ValueError(
f'Could not find the dataset: {dataset_name} in {experiment["annotations_path"]}'
)
modalities = experiment['modalities']
if modalities == 'same_as_train_set':
modalities = config['experiments']['train_set'][
'modalities']
if modality_name in modalities:
if 'column_name' in modalities[modality_name]:
try:
colname = modalities[modality_name][
'column_name']
dg.append_values(modality_name=modality_name,
values=annotations[colname])
except KeyError as e:
Console_UI().warn_user(
f'Got a key annotation exception for {colname}'
)
Console_UI().warn_user(
modalities[modality_name])
Console_UI().warn_user(annotations.columns)
raise e
except Exception as e:
Console_UI().warn_user(
f'Got an annotation exception for {colname}'
)
Console_UI().warn_user(
modalities[modality_name])
Console_UI().warn_user(annotations)
raise e
elif 'columns' in modalities[modality_name]:
for column_name in modalities[modality_name][
'columns']:
if isinstance(column_name, dict):
assert 'csv_name' in column_name, \
f'The column doesn\'t have the expected csv_name element, got: {column_name}'
column_name = column_name['csv_name']
if column_name not in annotations:
n = 3 if len(
annotations.columns) < 10 else ceil(
len(annotations.columns) / 3)
closest = get_close_matches(
word=column_name,
possibilities=annotations.columns,
n=n,
)
closest = ', '.join(closest)
raise IndexError(
f'The {column_name} from {modality_name} doesn\'t exist.'
+ f' Closest matching are: {closest}')
dg.append_values(
modality_name=modality_name,
values=annotations[column_name])
else:
raise IndexError(
f'Expected {modality_name} to have either columns or column_name defined'
)
import os
import Tkinter as tk
from Tkinter import *
import shutil
import pyglet
from file_manager import File_Manager
import Tkinter, Tkconstants, tkFileDialog
root = Tk()
root.minsize(300, 300)
number_of_folders = input("number_of_folders")
foo = File_Manager(number_of_folders)
foo.folder_name_and_extension()
foo.sort()
def get_graph_cfgs(self, graph_name):
graph_cfgs = File_Manager().read_graph_config(graph_name)
graph_cfgs = self.fix_experiment_modalities(graph_cfgs)
return graph_cfgs
def print_table(table):
for x in sorted(table, key=lambda leaf:leaf.name[0], reverse=True):
print x.name[0], ' - ', x.freq, ' - ', x.code
###########################################################################################
# MAIN PROGRAM
###########################################################################################
print "Program start."
img = get_image('lena_gray.tiff')
# img = get_image('c.jpg')
manager = File_Manager('test')
S, H = get_symbols_and_occurrences(img)
D = get_tuple_array(S, H)
t0 = time()
Tree, Code = get_huffman(D)
t1 = time()
#print_table(Code)
print '[Time] Huffman = ', t1 - t0
keys = [c.name[0] for c in Code]
values = [c.code for c in Code]
dicc = dict(zip(keys, values))
def run_scene(self, start_epoch=0):
logged_memory_usage = False
ui = Console_UI()
ui.overall_total_epochs = self.epochs
ui.overall_total_repeats = self.repeat
Global_Cfgs().set_forward_noise(
self.get_cfgs('forward_noise', default=0))
for r in range(0, self.repeat):
ui.overall_repeat = r
if (self.stochastic_weight_averaging and r > 0):
self.tasks[self.main_task].stochastic_weight_average()
for e in range(0, self.epochs):
ui.overall_epoch = e
if start_epoch > e + r * self.epochs:
Scene.iteration_counter += self.epoch_size
else:
for task in self.tasks.values():
task.update_learning_rate(self.get_learning_rate(e))
for _ in range(self.epoch_size):
for key, task in self.tasks.items():
if self.should_task_run(task_name=key, task=task):
task.step(
iteration_counter=Scene.iteration_counter,
scene_name=self.scene_name)
Scene.iteration_counter += 1
if logged_memory_usage is False:
for key in self.tasks.keys():
task = self.tasks[key]
memory_usage = task.get_memory_usage_profile()
File_Manager().write_usage_profile(
scene_name=self.scene_name,
task=key,
memory_usage=memory_usage,
)
ui.inform_user(
f'\n Memory usage for {self.scene_name}::{key}\n'
)
ui.inform_user(memory_usage)
logged_memory_usage = True
for task in self.tasks.values():
task.save(scene_name='last')
# Not really helping with just emptying cache - we need to add something more
# removing as this may be the cause for errors
# torch.cuda.empty_cache()
ui.reset_overall()
# Note that the evaluation happens after this step and therefore averaging may hur the performance
if self.stochastic_weight_averaging_last:
self.tasks[self.main_task].stochastic_weight_average()
for task in self.tasks.values():
task.save(scene_name='last')
for task in self.tasks.values():
task.validate(iteration_counter=Scene.iteration_counter,
scene_name=self.scene_name)
task.test(iteration_counter=Scene.iteration_counter,
scene_name=self.scene_name)
# Save all tasks before enterering the next scene
for task in self.tasks.values():
task.save(scene_name=self.scene_name)
[g.dropModelNetworks() for g in task.graphs.values()]
