def playlist_dl(playlist_id):
data=get_playlist(playlist_id)
print('Start Dowdloading {0}.'.format(data['name']))
path=create_dir(data)
write_info_json(path,data)
songs=get_songs_info(path,data)
global ID_WIDTH
ID_WIDTH=len(str(len(songs)))
print('Dowdloading songs...')
tqdm.get_lock()
with tqdm(total=len(songs),ncols=70) as pbar:
pool=ThreadPoolExecutor(max_workers=MAX_POOL)
tasks=[pool.submit(download_song,path,i+1,song) for i,song in enumerate(songs)]
for _ in as_completed(tasks):
pbar.update()
def outer_function(**config):
"""Outer function running inner function for each task in input dict"""
freeze_support() # for Windows support
tqdm.set_lock(RLock())
with concurrent.futures.ThreadPoolExecutor(initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(), ),
max_workers=3) as executor:
results_list = []
outer_loop_kwarg = {
'total': len(config['package']['tasks']),
'desc': 'Outer',
'ascii': True,
'position': len(config['package']['tasks']),
'leave': True
}
with tqdm(**outer_loop_kwarg) as out_progress:
futuresListComp = [
executor.submit(inner_function, **node)
for node in config['package']['tasks']
]
# Update after each completed task
for future in concurrent.futures.as_completed(futuresListComp):
out_progress.update()
results_list.append(future.result())
return results_list
def main():
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument("--datapath", type=str, default="data")
parser.add_argument("--data_type", type=str, default="train")
parser.add_argument("--pilot_version", type=int, choices=[1, 2], default=1)
parser.add_argument("--processes", type=int, default=4)
parser.add_argument("--data_nums", type=int, default=64)
parser.add_argument("--seed", type=int, default=43)
parser.add_argument("--mode", type=int, choices=[0, 1, 2], default=None)
parser.add_argument("--SNRdb", type=float, default=None)
parser.add_argument("--with_pure_y", action='store_true')
parser.add_argument("--debug", action='store_true')
args = parser.parse_args()
H, Htest = read_data(args.datapath)
using_H = H if args.data_type == "train" else Htest
generate_data_fix = partial(generate_data, args=args, H=using_H)
tqdm.set_lock(RLock())
with Pool(processes=args.processes,
initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(), )) as pool:
[
pool.map(generate_data_fix,
range(args.processes * i, args.processes * (i + 1)))
for i in range(args.data_nums // args.processes)
]
def _do_epoch(self, data_iter, is_train, batches_count, name=None):
self.on_epoch_begin(is_train, name, batches_count=batches_count)
progress_bar_class = ConsoleProgressBar
if self._use_tqdm:
try:
from tqdm import tqdm
tqdm.get_lock().locks = []
progress_bar_class = tqdm
except:
pass
with torch.autograd.set_grad_enabled(is_train):
with progress_bar_class(total=batches_count) as progress_bar:
try:
for _ in range(batches_count):
batch = next(data_iter)
batch_progress = self.on_batch(batch)
progress_bar.update()
progress_bar.set_description(batch_progress)
except StopIteration:
pass
epoch_progress = self.on_epoch_end()
progress_bar.set_description(epoch_progress)
progress_bar.refresh()
def scrape():
global filename
start_time = datetime.now()
# Building the filename
filename = str(filename).replace("$DATE", start_time.strftime("%Y%m%d%H%M%S"))
search = str(args.search).replace(" ", "")
if len(search) > 10:
search = search[0:9]
filename = str(filename).replace("$SEARCH", search)
func_args = []
stats_dict = {}
if args.engines and len(args.engines) > 0:
eng = args.engines[0]
for e in eng:
try:
if not (args.exclude and len(args.exclude) > 0 and e in args.exclude[0]):
func_args.append("{}:{}".format(e, args.search))
stats_dict[e] = 0
except KeyError:
print("Error: search engine {} not in the list of supported engines".format(e))
else:
for e in supported_engines.keys():
if not (args.exclude and len(args.exclude) > 0 and e in args.exclude[0]):
func_args.append("{}:{}".format(e, args.search))
stats_dict[e] = 0
# Doing multiprocessing
units = min((cpu_count() - 1), len(func_args))
if args.mp_units and args.mp_units > 0:
units = min(args.mp_units, len(func_args))
print("search.py started with {} processing units...".format(units))
freeze_support()
results = {}
with Pool(units, initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),)) as p:
results_map = p.map(run_method, func_args)
results = reduce(lambda a, b: a + b if b is not None else a, results_map)
stop_time = datetime.now()
if not args.continuous_write:
with open(filename, 'w', newline='') as csv_file:
csv_writer = csv.writer(csv_file, delimiter=field_delim, quoting=csv.QUOTE_ALL)
for r in results:
write_to_csv(csv_writer, r)
total = 0
print("\nReport:")
print(" Execution time: %s seconds" % (stop_time - start_time))
print(" Results per engine:")
for r in results:
stats_dict[r['engine']] += 1
for s in stats_dict:
n = stats_dict[s]
print(" {}: {}".format(s, str(n)))
total += n
print(" Total: {} links written to {}".format(str(total), filename))
def proc_2():
cp = CustomPool()
with cp.Pool(n_process=4, initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),)) as p:
for result in tqdm(p.imap_unordered(progresser_2, range(10)), total=10):
pass
cp.update()
def main(args):
with open(args.vectors_file, mode="r") as fp:
vectors_file = json.load(fp)
vectors = vectors_file["vectors"]
vectors_dir = os.path.dirname(args.vectors_file)
jobs_list = list(combinations(vectors.keys(), 2))
splitted_jobs = split_jobs(jobs_list, args.threads)
logging.debug("Jobs count: {}".format(len(jobs_list)))
logging.debug(" For each worker: {}".format([len(splitted) for splitted in splitted_jobs]))
vectorbase_path = os.path.normpath(os.path.join(vectors_dir, vectors_file["base"]["path"]))
if not os.path.isfile(vectorbase_path):
logging.error("VectorBase file '{}' does not exist".format(vectorbase_path))
sys.exit(1)
scores = {}
freeze_support() # for Windows support
with Pool(args.threads, initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),)) as pool:
api_vector = ApiVector(vectorbase_path)
logging.info("Calculating the scores...")
results = []
for i in range(args.threads):
task_args = (i, args.threads, api_vector, splitted_jobs[i], vectors, args.verbose)
results.append(pool.apply_async(worker_function, args=task_args))
pool.close()
pool.join()
logging.info(" COMPLETED")
logging.info("Merging the results... ")
scores = merge_dictionaries([res.get() for res in results])
logging.info(" COMPLETED")
out_dir = os.path.join(os.path.dirname(vectors_dir), "scores")
base = vectors_file["base"]
if args.out_file is None:
args.out_file = os.path.join(out_dir, "scores_{}_{}_{}.json".format(base["imports_type"],
base["weights"],
base["size"]))
if not os.path.exists(os.path.dirname(args.out_file)):
try:
os.makedirs(os.path.dirname(args.out_file))
except OSError as exc: # Guard against race condition
if exc.errno != errno.EEXIST:
raise
dataset_path = os.path.normpath(os.path.join(vectors_dir, vectors_file["dataset"]))
scores_dict = {
"dataset": os.path.relpath(dataset_path, out_dir),
"base": vectors_file["base"],
"scores": scores
}
scores_dict["base"]["path"] = os.path.relpath(vectorbase_path, out_dir)
with open(args.out_file, mode="w") as opf:
logging.info("Dumping scores to '{}'... ".format(os.path.basename(args.out_file)))
json.dump(scores_dict, opf, indent=4)
logging.info(" COMPLETED")
def prog_map(elms, f, desc="Synth", chunksize=1,procs=8,order=True):
with tqdm(elms, desc=desc) as t:
with multiprocessing.Pool(procs, initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(),)) as p:
if (order):
pool = list(p.imap(f, t, chunksize=chunksize))
else:
pool = list(p.imap_unordered(f, t, chunksize=chunksize))
return pool
def dfu_flash(dfu, dfu_file, queue, pos=0):
'''Flash a list of DFU devices with the given file'''
snum = dfu.get_string(dfu.dev.iSerialNumber)
# Clear left-over errors
if dfu.get_status()[1] == kidfu.DfuState.DFU_ERROR:
dfu.clear_status()
# Flash
blocks = [
dfu_file.data[i:i + 64] for i in range(0, len(dfu_file.data), 64)
]
with tqdm.get_lock():
progress = tqdm(total=len(blocks),
unit='B',
unit_scale=64,
miniters=0,
desc=colorize(snum, colorama.Fore.CYAN),
position=pos,
dynamic_ncols=True,
leave=True,
smoothing=0)
for bnum, block in enumerate(blocks):
try:
dfu.write(bnum, block)
status = dfu.wait_while_state(kidfu.DfuState.DFU_DOWNLOAD_BUSY)
if status[1] != kidfu.DfuState.DFU_DOWNLOAD_IDLE:
queue.put('%s: Error %d' % (snum, status[1]))
return
except usb.core.USBError:
queue.put('%s: USB error' % snum)
return
with tqdm.get_lock():
progress.update(1)
with tqdm.get_lock():
progress.refresh()
progress.close()
dfu.leave()
status = dfu.get_status()
if status[1] == kidfu.DfuState.DFU_MANIFEST_SYNC:
queue.put('%s: OK' % snum)
return
queue.put('%s: Error finish' % snum)
def test_multiprocessing_1():
hyper = Hyperactive(
distribution={
"multiprocessing": {
"initializer": tqdm.set_lock,
"initargs": (tqdm.get_lock(), ),
}
})
hyper.add_search(objective_function, search_space, n_iter=15, n_jobs=2)
hyper.run()
def get_songs_info(path,data):
path_json=os.path.join(path,'mini_info.json')
if os.path.exists(path_json):
with open(path_json) as f:
return json.load(f)
songs=[]
print('Geting songs information...')
tqdm.get_lock()
with tqdm(total=len(data['tracks']),ncols=70) as pbar:
pool=ThreadPoolExecutor(max_workers=MAX_POOL)
tasks=[pool.submit(get_one_song_info,songs,i,x) for i,x in enumerate(data['tracks'])]
for state in as_completed(tasks):
if state._result:
pbar.update()
songs=sorted(songs,key=lambda x: x[0])
songs=[x[1] for x in songs]
with open(path_json,'w') as f:
json.dump(songs,f)
return songs
def parallel_ilr_inference(nb_jobs=50, **kwargs):
kwargs_list = []
for n in range(nb_jobs):
kwargs['seed'] = n
kwargs_list.append(kwargs.copy())
with Pool(processes=min(nb_jobs, nb_cores),
initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(), )) as p:
res = p.map(_job, kwargs_list)
return res
def spawn_processes(init_fn, read_fn, write_fn, num_readers=1, num_writers=1):
"""Start readers and writers."""
tqdm.set_lock(RLock())
write_q = Queue()
write_fn = partial(write_fn, lock=tqdm.get_lock(), write_q=write_q)
writers = [
Process(daemon=True, target=partial(write_fn, pos=i))
for i in range(num_writers, 0, -1)
]
read_q = Queue()
read_fn = partial(read_fn, lock=tqdm.get_lock(), read_q=read_q, write_q=write_q)
readers = [
Process(daemon=True, target=partial(read_fn, pos=i))
for i in range(num_readers + 1, num_writers, -1)
]
for p in readers + writers:
p.start()
init_fn(read_q)
notify_and_join(read_q, readers)
notify_and_join(write_q, writers)
def scp_operation(args):
credentials = Credentials()
targets = args.hostname.split(",")
tqdm.set_lock(RLock())
with Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(), )) as pool:
pool.starmap(
scp_process,
zip(
targets,
repeat(credentials),
repeat(args.filename),
repeat(args.dst_file_path),
list(range(len(targets))),
),
)
def parallel_ilr_inference(nb_jobs=50, **kwargs):
kwargs_list = []
for n in range(nb_jobs):
_kwargs = {'seed': kwargs['arguments'].seed,
'train_input': kwargs['train_input'][n],
'train_target': kwargs['train_target'][n],
'arguments': kwargs['arguments']}
kwargs_list.append(_kwargs)
with Pool(processes=min(nb_jobs, nb_cores),
initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(),)) as p:
res = p.map(_job, kwargs_list)
return res
def process_archives(archive_links: List[str], target_dir: str,
processes_number: int) -> None:
"""
Download tar files and untar them to target directory
:param archive_links: List of tar files links
:param target_dir: Target directory
:param processes_number: Number of processes which will be used in multiprocessing download
:return: None
"""
os.makedirs(target_dir, exist_ok=True)
pool = Pool(processes=processes_number,
initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(), ))
process_archive_with_arg = partial(process_archive, target_dir)
for _ in tqdm(pool.imap(process_archive_with_arg,
enumerate(archive_links)),
desc="process archives",
total=len(archive_links),
position=0):
pass
def __init__(
self,
verbosity=["progress_bar", "print_results", "print_times"],
distribution={
"multiprocessing": {
"initializer": tqdm.set_lock,
"initargs": (tqdm.get_lock(),),
}
},
):
if verbosity is False:
verbosity = []
self.verbosity = verbosity
self.distribution = distribution
self.search_ids = []
self.process_infos = {}
self.objFunc2results = {}
self.search_id2results = {}
def __init__(
self,
verbosity=["progress_bar", "print_results", "print_times"],
distribution={
"multiprocessing": {
"initializer": tqdm.set_lock,
"initargs": (tqdm.get_lock(), ),
}
},
n_processes="auto",
):
super().__init__()
if verbosity is False:
verbosity = []
self.verbosity = verbosity
self.distribution = distribution
self.n_processes = n_processes
self.search_ids = []
self.process_infos = {}
self.progress_boards = {}
def compute_by_window(imgs, func, window_size=16, step=2, dst_dtype=np.float32, n_worker=12):
"""
画像を一部を切り取り、func 関数で行った計算結果を
返却する
Parameters
----------
imgs : numpy.ndarray or tuple of numpy.ndarray
入力画像
tuple で複数画像を与える場合、各画像に対して
同じ領域を切り取り、処理を行うため、各画像の
縦、横サイズは一致している必要がある
func : callable object
切り取った画像の一部に対して何らかの計算を行う
関数。引数として画像の一部が渡される。
window_size : int or tuple of int
画像を切り取るサイズ。
int を指定した場合は、縦横同じサイズで切り取る。
tuple(int, int) を指定した場合は、縦横で異なったサイズ
で切り取り、指定する順序は ndarray の次元に対応する
step : int or tuple of int
切り取り間隔
int を指定した場合は、縦横同じ間隔を開けて処理をする
tuple(int, int) を指定した場合は、縦横で異なった間隔
を開けて処理を行い、指定する順序は ndarray の次元に
対応する
dst_dtype : type, default numpy.float32
返却値のデータ型
n_worker : int, default 4
並列するプロセス数
Returns
-------
numpy.ndarray
各切り取り画像に対する処理結果の行列
"""
# TYPE ASSERTION
TYPE_ASSERT(imgs, [np.ndarray, tuple])
TYPE_ASSERT(window_size, [int, tuple])
TYPE_ASSERT(step, [int, tuple])
TYPE_ASSERT(dst_dtype, type)
TYPE_ASSERT(n_worker, int)
if isinstance(imgs, np.ndarray):
imgs = tuple([imgs])
for img in imgs:
TYPE_ASSERT(img, np.ndarray)
for i in range(len(imgs) - 1):
SAME_SHAPE_ASSERT(imgs[i], imgs[i + 1])
n_imgs = len(imgs)
height, width = imgs[0].shape[:2]
assert callable(func) and n_args(func) >= n_imgs, \
"argument 'func' must be callable object which has {0} argument at least. \n".format(n_imgs) + \
" ( num of argumets of 'func' depends on argument 'imgs')"
if isinstance(step, int):
step = tuple([step] * 2)
if isinstance(window_size, int):
window_size = tuple([window_size] * 2)
s_i, s_j = step
w_w, w_h = window_size
results_shape = ceil(height / s_i), ceil(width / s_j)
# Add padding to input images
eprint("Add padding ... ")
imgs = [
np.pad(
img,
pad_width=[
tuple([w_w // 2]),
tuple([w_h // 2]),
] + [] if img.ndim == 2 else [tuple([0])],
mode="constant",
constant_values=0
)
for img in imgs
]
if n_worker == 1:
results = np.ndarray(results_shape, dtype=dst_dtype)
for ii, i in tqdm(enumerate(range(w_h // 2, height + w_h // 2, s_i)), total=results_shape[0]):
for jj, j in tqdm(enumerate(range(w_w // 2, width + w_w // 2, s_j)), total=results_shape[1], leave=False):
rois = [
img[
get_window_rect(
img.shape,
center=(j, i),
wnd_size=(w_w, w_h),
ret_type="slice"
)
]
for img in imgs
]
results[ii][jj] = func(*rois)
else:
global _func
global _callee
_func = func
def _callee(_imgs, _func, _width, _s_j, _w_w, _n_loop):
_worker_id = current_process()._identity[0]
_desc = f"Worker #{_worker_id:3d}"
_results = list()
for jj, j in tqdm(enumerate(range(_w_w // 2, _width + _w_w // 2, _s_j)), total=_n_loop, desc=_desc,
position=_worker_id,
leave=False):
_rois = [
# _roi[:, j:j + _w_w]
_roi[
get_window_rect(
_roi.shape,
center=(j, -1),
wnd_size=(_w_w, -1),
ret_type="slice"
)
]
for _roi in _imgs
]
_results.append(_func(*_rois))
return _results
progress_bar = tqdm(total=results_shape[0], position=0)
def _update_progressbar(arg):
progress_bar.update()
cp = CustomPool()
pool = cp.Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),))
results = list()
for ii, i in enumerate(range(w_h // 2, height + w_h // 2, s_i)):
rois = [
img[
get_window_rect(
img.shape,
center=(-1, i),
wnd_size=(-1, w_h),
ret_type="slice"
)
]
for img in imgs
]
results.append(
pool.apply_async(
_callee,
args=(rois, func, width, s_j, w_h, results_shape[1]),
callback=_update_progressbar
)
)
pool.close()
pool.join()
cp.update()
results = np.array(
[result.get() for result in results],
dtype=dst_dtype
)
return results
# we think we know about other bars (currently only py3 threading)
if n == 6:
tqdm.write("n == 6 completed")
if __name__ == '__main__':
freeze_support() # for Windows support
L = list(range(NUM_SUBITERS))[::-1]
print("Manual nesting")
for i in trange(16, desc="1"):
for _ in trange(16, desc="2 @ %d" % i, leave=i % 2):
sleep(0.01)
print("Multi-processing")
p = Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),))
p.map(progresser, L)
# unfortunately need ncols
# to print spaces over leftover multi-processing bars (#796)
with tqdm(leave=False) as t:
ncols = t.ncols or 80
print(("{msg:<{ncols}}").format(msg="Multi-threading", ncols=ncols))
# explicitly set just threading lock for nonblocking progress
tqdm.set_lock(RLock())
with ThreadPoolExecutor() as p:
progresser_thread = partial(
progresser, write_safe=not PY2, blocking=False)
p.map(progresser_thread, L)
delimiter=',',
fieldnames=OUTPUT_FIELDS)
output_csv.writerow(dict((fn, fn) for fn in OUTPUT_FIELDS))
# Gather all rows into memory
all_input_rows = [input_row for input_row in input_csv]
# Split rows into NUM_CORES chunks for parallel processing
input_chunks = numpy.array_split(numpy.array(all_input_rows), NUM_CORES)
input_chunks_with_index = [(index, chunk)
for index, chunk in enumerate(input_chunks)]
# Create pool of workers
pool = mp.Pool(initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(), ),
processes=NUM_CORES)
# Apply transformation in parallel
print("Starting transformation with %s workers..." % NUM_CORES)
output_chunks = pool.starmap(batch_tranform_to_rayyan,
input_chunks_with_index)
# Wrap up workers
pool.close()
pool.join()
# Write output to file
for chunk in output_chunks:
for row in chunk:
output_csv.writerow(row)
def classify(self):
"""
エッジ画素分類を行う
- テンプレートマッチングにより
「端点」「分岐点」を探索
- テンプレートにマッチしなかったエッジ画素は
「通過点」とする
Returns
-------
numpy.ndarray
エッジ画素分類結果
"""
self.check_image()
BG, FG = self.BG, self.FG
k_size = self.K_SIZE
logger = self.logger
img = self.img
# 幅 1 のパディングを追加
img = np.pad(img,
pad_width=k_size // 2,
mode="constant",
constant_values=BG)
height, width = img.shape[:2]
progress_bar = tqdm(total=height, position=0)
def _update_progressbar(arg):
progress_bar.update()
cp = CustomPool()
pool = cp.Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(), ))
results = list()
# 全画素ループ (tqdm で進捗可視化)
for i in trange((k_size // 2),
height - (k_size // 2),
desc="Height",
leave=False):
roi = img[i:i + k_size, :]
results.append(
pool.apply_async(self._classify_pixel,
args=(roi, ),
callback=_update_progressbar))
pool.close()
pool.join()
cp.update()
self.classified = np.array([result.get() for result in results],
dtype=self.classified.dtype)
if logger:
logger.logging_img(self.classified, "classified")
logger.logging_img(self.get_as_image(), "classified_visualized")
return self.classified
def render_parallel(num_jobs, scene, frame_window=None, **kwargs):
import functools
import multiprocessing
from multiprocessing import RLock
import h5py
from tqdm import tqdm
from gwpv.scene_configuration import animate, parse_as
logger = logging.getLogger(__name__)
# Infer frame window if needed
if "FreezeTime" in scene["Animation"]:
frame_window = (0, 1)
elif frame_window is None:
if "Crop" in scene["Animation"]:
max_animation_length = (scene["Animation"]["Crop"][1] -
scene["Animation"]["Crop"][0])
else:
waveform_file_and_subfile = parse_as.file_and_subfile(
scene["Datasources"]["Waveform"])
with h5py.File(waveform_file_and_subfile[0], "r") as waveform_file:
waveform_times = waveform_file[
waveform_file_and_subfile[1]]["Y_l2_m2.dat"][:, 0]
max_animation_length = waveform_times[-1] - waveform_times[0]
logger.debug(
f"Inferred max. animation length {max_animation_length}M"
" from waveform data.")
frame_window = (
0,
animate.num_frames(
max_animation_length=max_animation_length,
animation_speed=scene["Animation"]["Speed"],
frame_rate=scene["Animation"]["FrameRate"],
),
)
logger.debug(f"Inferred total frame window: {frame_window}")
num_frames = frame_window[1] - frame_window[0]
frames_per_job = int(num_frames / num_jobs)
extra_frames = num_frames % num_jobs
logger.debug(f"Using {num_jobs} jobs with {frames_per_job} frames per job"
f" ({extra_frames} jobs render an additional frame).")
frame_windows = []
distributed_frames = frame_window[0]
for i in range(num_jobs):
frames_this_job = frames_per_job + (1 if i < extra_frames else 0)
frame_windows.append(
(distributed_frames, distributed_frames + frames_this_job))
distributed_frames += frames_this_job
logger.debug(f"Frame windows: {frame_windows}")
tqdm.set_lock(RLock())
pool = multiprocessing.Pool(num_jobs,
initializer=tqdm.set_lock,
initargs=(tqdm.get_lock(), ))
from gwpv.render.frames import _render_frame_window
render_frame_window = functools.partial(_render_frame_window,
scene=scene,
**kwargs)
pool.starmap(render_frame_window, enumerate(frame_windows))
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
import pickle, sys, random, time, logging, argparse
from fnmatch import fnmatch
from copy import deepcopy
from typing import List, Tuple, Optional, Text, FrozenSet
from abc import abstractmethod
from operator import itemgetter
from collections import defaultdict
from itertools import chain
from tqdm import tqdm
# work around pypy bug https://bitbucket.org/pypy/pypy/issues/2953/deadlock
tqdm.get_lock().locks = []
import yaml
from .layout import defaultLayouts, ButtonCombination, Layer, KeyboardLayout, GenericLayout
from .carpalx import Carpalx, models, ModelParams
from .writer import Writer
from .util import first
from .keyboard import defaultKeyboards, LetterButton
class Annealer:
"""
Simulated annealing.
Override .mutate() to suit your needs. Uses exponential cooling (10^(-progress*factor))
def _initialize(proc_name):
tqdm.set_lock(tqdm.get_lock())
current_process().name = proc_name
def log_progress(items, prefix=None, total=None):
# https://github.com/tqdm/tqdm/issues/461#issuecomment-334343230
tqdm.get_lock().locks = []
return tqdm(items, desc=prefix, total=total)
if __name__ == "__main__":
input_csv = csv.DictReader(open(sys.argv[1], 'r', encoding='utf-8', errors='ignore'), delimiter=',')
output_csv = csv.DictWriter(open(sys.argv[2], "w+"), delimiter=',', fieldnames=OUTPUT_FIELDS)
output_csv.writerow(dict((fn, fn) for fn in OUTPUT_FIELDS))
# Gather all rows into memory
all_input_rows = [input_row for input_row in input_csv]
# Split rows into NUM_CORES chunks for parallel processing
input_chunks = numpy.array_split(numpy.array(all_input_rows), NUM_CORES)
input_chunks_with_index = [(index, chunk) for index, chunk in enumerate(input_chunks)]
# Create pool of workers
pool = mp.Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),), processes=NUM_CORES)
# Apply transformation in parallel
print("Starting transformation with %s workers..." % NUM_CORES)
output_chunks = pool.starmap(batch_tranform_to_rayyan, input_chunks_with_index)
# Wrap up workers
pool.close()
pool.join()
# Write output to file
for chunk in output_chunks:
for row in chunk:
output_csv.writerow(row)
print("Complete.")
def find_color_threshold_in_hsv(self, img, ground_truth, precision=10):
"""
HSV 色空間における色閾値探索
- RGB → HSV 変換 を行う
- HSV の各チャンネルで閾値処理を行い統合する
- 正解データを用いて精度評価を行う
Parameters
----------
img : numpy.ndarray
入力画像 (8-Bit RGB カラー)
ground_truth : numpy.ndarray
正解データ (1-Bit)
precision : int
閾値計算の精度
Returns
-------
reasonable_params : dict
F値が最も高くなったときの閾値
result : numpy.ndarray
その際の閾値処理結果画像 (1-Bit 2値画像)
Notes
-----
`ground_truth`:
- 1-Bit (bool 型) 2値画像
- 黒：背景、白：被害領域
`precision`:
- precision=N のとき、H, S, V の各チャンネルに対し
2N ずつにパーセンタイル分割を行う
"""
global _worker_find_color_threshold_in_hsv
# Worker methods executed parallel
@worker_exception_raisable
def _worker_find_color_threshold_in_hsv(_img, _masked, _q_h, _q_s):
# Value used in tqdm
_worker_id = current_process()._identity[0]
_desc = f"Worker #{_worker_id:3d}"
# Unpack arguments
_q_h_low, _q_h_high = _q_h
_q_s_low, _q_s_high = _q_s
# Split image to each channne
_img_h, _img_s, _img_v = [_img[:, :, i] for i in range(3)]
_masked_h, _masked_s, _masked_v = [_masked[:, i] for i in range(3)]
# Initialize variables
reasonable_params = {
"Score": {
"F Score": -1,
},
"Range": -1
}
# Find thresholds
for _q_v_low, _q_v_high in tqdm(list(
product(np.linspace(50 / precision, 50, precision),
repeat=2)),
desc=_desc,
position=_worker_id,
leave=False):
# Generate result
_h_min, _h_max = self._in_range_percentile(
_masked_h, (_q_h_low, _q_h_high))
_s_min, _s_max = self._in_range_percentile(
_masked_s, (_q_s_low, _q_s_high))
_v_min, _v_max = self._in_range_percentile(
_masked_v, (_q_v_low, _q_v_high))
_result = (((_h_min <= _img_h) & (_img_h <= _h_max)) &
((_s_min <= _img_s) & (_img_s <= _s_max)) &
((_v_min <= _img_v) & (_img_v <= _v_max)))
# Calculate score
_cm, _metrics = evaluation_by_confusion_matrix(
_result, ground_truth)
# Update reasonable_params
if _metrics["F Score"] > reasonable_params["Score"]["F Score"]:
reasonable_params = {
"Score": _metrics,
"Confusion Matrix": _cm,
"Range": {
"H": (_h_min, _h_max, _q_h_low, _q_h_high),
"S": (_s_min, _s_max, _q_s_low, _q_s_high),
"V": (_v_min, _v_max, _q_v_low, _q_v_high),
}
}
return reasonable_params
# Check arguments
NDARRAY_ASSERT(img, ndim=3, dtype=np.uint8)
NDARRAY_ASSERT(ground_truth, ndim=2, dtype=np.bool)
SAME_SHAPE_ASSERT(img, ground_truth, ignore_ndim=True)
# Convert RGB -> HSV
img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
# `masked`: `img` masked by `ground_truth`
masked = img[ground_truth]
# Percentile Split
Q = list(product(np.linspace(50 / precision, 50, precision), repeat=4))
# `progress_bar`: whole progress bar
progress_bar = tqdm(total=len(Q), position=0)
def _update_progressbar(arg):
progress_bar.update()
# Initialize process pool
cp = CustomPool()
pool = cp.Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(), ))
results = list()
# Multi-Processing !
for q_h_low, q_h_high, q_s_low, q_s_high in Q:
results.append(
pool.apply_async(_worker_find_color_threshold_in_hsv,
args=(img, masked, (q_h_low, q_h_high),
(q_s_low, q_s_high)),
callback=_update_progressbar))
pool.close()
pool.join()
cp.update()
# Resolve results
try:
results = [result.get() for result in results]
except Exception as e:
print(e)
# Get result whose F-Score is max in results
reasonable_params = max(results, key=lambda e: e["Score"]["F Score"])
img_h, img_s, img_v = [img[:, :, i] for i in range(3)]
h_min, h_max, _, _ = reasonable_params["Range"]["H"]
s_min, s_max, _, _ = reasonable_params["Range"]["S"]
v_min, v_max, _, _ = reasonable_params["Range"]["V"]
# Generate image using reasonable thresholds
result = (((h_min <= img_h) & (img_h <= h_max)) &
((s_min <= img_s) & (img_s <= s_max)) & ((v_min <= img_v) &
(img_v <= v_max)))
# Logging
if self.logger:
self.logger.logging_dict(reasonable_params,
"color_thresholds_in_hsv",
sub_path=self.logger_sub_path)
self.logger.logging_img(result,
"meanshift_thresholded",
sub_path=self.logger_sub_path)
return reasonable_params, result
def process_collection(collection):
fid = collection["foreign_id"]
fid = fid.replace("/", "")
fname = f"./dataset_components/{fid}.json"
if os.path.exists(fname):
return
try:
components = calculate_components(collection)
except AlephException as e:
print(f"Aleph Error: {fid}: {e}")
return
with open(fname, "w+") as fd:
data = {
"components_histogram": dict(components),
"collection": collection,
}
fd.write(json.dumps(data))
if __name__ == "__main__":
init_aleph()
collections = api.filter_collections("*")
N = collections.result["total"]
tqdm.set_lock(mp.RLock())
with mp.Pool(processes=4, initializer=init_aleph, initargs=(tqdm.get_lock(),)) as p:
results = p.imap_unordered(process_collection, collections, chunksize=32)
for _ in tqdm(results, total=N, position=0):
pass
def find_ms_params(n):
file_name = f"aerial_roi{n}.png"
src = imread_with_error(
join(ROOT_DIR_SRC, file_name)
)
ans = imread_with_error(
join(ROOT_DIR_ANS, file_name)
)
ms_params = sum([
[
{
"spatial_radius": sp,
"range_radius": sr,
"min_density": 0
}
for sr in np.arange(SR_RANGE[0], SR_RANGE[0]+SR_RANGE[1], SR_RANGE[2])
]
for sp in np.arange(SP_RANGE[0], SP_RANGE[0]+SP_RANGE[1], SP_RANGE[2])
], [])
progress_bar = tqdm(total=len(ms_params), position=0)
def _update_progressbar(arg):
progress_bar.update()
cp = CustomPool()
pool = cp.Pool(n_process=6, initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),))
results = list()
for params in ms_params:
results.append(
pool.apply_async(
func_worker,
args=(src, ),
kwds=params,
callback=_update_progressbar
)
)
pool.close()
pool.join()
cp.update()
results = [result.get() for result in results]
results = sorted(
[
(
sp,
sr,
np.sum(
np.abs(segmented - ans)
)
)
for segmented, sp, sr in results
],
key=lambda e: e[0]
)
pprint(results)
with open(f"tmp/find_ms_params_{n}.csv", "wt") as f:
f.write("spatial_radius, range_radius, n_diffs\n")
for result in results:
f.write(", ".join([ str(x) for x in result]) + "\n")
return results
