def process():
session = getUserInSession()
ext = request.form.get('ext')
#dropping table with the same number as the current session has if it somehow exists
engine = create_engine(db_url)
sql = text('DROP TABLE IF EXISTS table_' + session + ';')
engine.execute(sql)
#calling tracker from cmd that may be not correct (perhaps having the entire tracker code as an imported module would be better)
dir_path = os.path.dirname(os.path.realpath(__file__))
os.chdir(dir_path)
os.system(
"conda activate project & python object_tracker.py --video static/" +
session + "." + ext + " --output data/video/output.mp4")
#calling text detector
start.main(project_path + "static/" + session + "/",
project_path + "static/" + session + "/")
#calling text recogniser, 2nd parameter helps us to get db name etc
demo.main(project_path + "static/" + session + "/",
len(project_path) + 7, db_url)
#reading table and writing data to html
df = read_table(session, engine)
write_html('templates/' + session + '.html', df)
df.to_excel('static/' + session + '/results.xlsx')
return render_template(session + ".html")
def run_demo(picked_datasets_index):
# First clear some folders
remove_folder_content(demo_input_path)
remove_folder_content(demo_output_path)
# Now pick some random images from dataset
random_files = []
folder_path = datasets[picked_datasets_index]
for _ in range(files_to_pick):
random_files.append(folder_path + "/" +
random.choice(os.listdir(folder_path)))
for random_file in random_files:
shutil.copy2(random_file, demo_input_path)
# Now generate names for them
generate_demo_file_names.main()
import demo
demo.main()
# Now save results with renamed name to results folder
results_folder = '../results'
N = 10
file_index = 0
for file in os.listdir(demo_output_path):
# 0 - gt
# 1 - black and white
# 2 - out
file_extension = file.split('.')[-1]
if file_index % 3 == 0:
dataset_name = datasets[picked_datasets_index].split('/')[-1]
random_name = ''.join(
random.choices(string.ascii_uppercase + string.digits, k=N))
gt = f"{random_name}_gt.{file_extension}"
bandw = f"{random_name}_black_and_white.{file_extension}"
out = f"{random_name}_out.{file_extension}"
shutil.copy2(f"{demo_output_path}/{file}",
f"{results_folder}/{dataset_name}/{gt}")
elif file_index % 3 == 1:
# Black and white
shutil.copy2(f"{demo_output_path}/{file}",
f"{results_folder}/{dataset_name}/{bandw}")
else:
# out
shutil.copy2(f"{demo_output_path}/{file}",
f"{results_folder}/{dataset_name}/{out}")
file_index += 1
def models2df(model_names, opt, save_dir=None):
"""
Applies provided models to provided stimuli and outputs network activation
data as a DataFrame for further processing.
"""
activation_data = []
columns = [
"StimulusName", "ImageMode", "ModelName", "Region", "Activations"
]
for img_mode in opt.feeding_modes:
stim_source, mode = str.split(img_mode, "-")
img_paths = []
if stim_source == "original":
img_dir = opt.stimuli_dir
img_paths = sorted(os.listdir(img_dir))
elif stim_source == "rendered":
if os.path.exists(opt.stimuli_dir + '-trunk-centric'):
img_dir = opt.stimuli_dir + "-trunk-centric"
else:
img_dir = opt.stimuli_dir + "-trunk-centric-uncentered"
img_paths = sorted(os.listdir(img_dir))
img_paths = [path for path in img_paths if "-" not in path]
else:
print(
"Source in image mode {} is not implemented.".format(img_mode))
# loop through all stimuli/images in this mode
make_path('tmp')
for img_path in img_paths:
stimulus_name = str.split(os.path.basename(img_path), ".")[0]
# loop through image modes (e.g. "raw" or "cropped")
if mode == "raw":
img_path = os.path.join(img_dir, img_path)
elif mode == "cropped":
img_path = os.path.join(img_dir + '-cropped', img_path)
# elif mode == "attended":
# img = attend_crop_img(img, attend_size=opt.attend_size)
else:
print("Mode in image mode {} is not implemented.".format(mode))
# loop through models
for model_name in model_names:
print('\nComputing output for model {} and image {}.\n'.format(
model_name, stimulus_name))
output = main(img_path, model_type=model_name)
# loop through labels
for model_region in opt.model_regions[model_name]:
out = output[model_region].flatten()
activation_data.append([
stimulus_name, img_mode, model_name, model_region, out
])
return pd.DataFrame(activation_data, columns=columns)
def main(file_list = ["PID026.JPG"]):
from time import clock
predicted_symbols, classes, size = Dectection.main(file_list, DEBUG)
ret_list = []
t_start = clock()
for item in zip(predicted_symbols, file_list):
keys = item[0].keys()
img = Image.open(item[1])
coordinate_list = []
#draw = ImageDraw.Draw(img)eog
if 0 in keys:
for bound in item[0][0]: #Valve
symbol_dict = {}
coordinate = bound[0]
symbol_dict['coord'] = coordinate
symbol_dict['type'] = 0
img_cv = PIL2CV(img.crop(coordinate))
#Feature_matching(img_cv)
symbol_dict['detail'] = Specific_Valve(img_cv)
if symbol_dict['detail'] is None:
symbol_dict['detail'] = "None"
coordinate_list.append(symbol_dict)
if 2 in keys:
for bound in item[0][2]:
symbol_dict = {}
coordinate = bound[0]
symbol_dict['coord'] = coordinate
symbol_dict['type'] = 2
symbol_dict['detail'] = "None"
coordinate_list.append(symbol_dict)
if 1 in keys: #
for bound in item[0][1]:
symbol_dict = {}
coordinate = bound[0]
symbol_dict['coord'] = coordinate
symbol_dict['type'] = 1
for i in range(9, 3, -1):
temp = []
image_cv = convert_PIL_cv(img,coordinate,i)
OCR_str = pytesseract.image_to_string(image_cv, config=tess_CONFIG)
if len(OCR_str) != 1:
OCR_str = OCR_str.replace(' ','').split("\n")
temp = _OCR_string(OCR_str[0])
if len(OCR_str) != 1:
temp += _OCR_digit(OCR_str[1])
if len(temp) !=0:
break
symbol_dict['detail'] = temp
coordinate_list.append(symbol_dict)
ret_list.append(coordinate_list)
print("Execution Time : {}".format(clock() - t_start))
return ret_list, classes
def main():
"""Starts the application and creates a virtualenv in case of failure."""
_dir, _script = osp.split(osp.abspath(__file__))
_venv = osp.join(_dir, 'venv')
_bin = osp.join(_venv, 'Scripts' if sys.platform == 'win32' else 'bin')
_py = osp.join(_bin, 'python')
try:
# We try to import our application module
import demo
except ImportError:
# Our application fails to import dependencies
if not osp.exists(_venv):
# Creates a virtual environment and installs
# dependencies using pip
venv.create(_venv, with_pip=True)
pip.main(['install', '-r', 'requirements.txt', '--prefix', _venv])
# Execute the script with the venv python executable
subprocess.run([_py, _script])
else:
demo.main()
def stopRecording(self):
self.out.release()
self.capture = not self.capture
self.btn_snapshot["text"] = "Record"
self.class_label["text"] = "Loading label"
# self.playSound(self.class_label["text"])
print("Should extract")
extract_files.extract_file()
print("extraction done")
prediction = demo.main()
self.playSound(prediction)
self.class_label["text"] = prediction
def main(_):
trainig_folder = setup_logging_and_exp_folder()
FLAGS.cuda = FLAGS.cuda and torch.cuda.is_available()
logging.info('Use Cuda: {}'.format(FLAGS.cuda))
logging.info('Current git SHA: ' + CURR_VERSION)
# save options
fpath = os.path.join(trainig_folder, 'flagfile')
with open(fpath, 'w') as f:
f.write(FLAGS.flags_into_string())
if FLAGS.mode == 'IL':
IL.run(training_folder=trainig_folder)
elif FLAGS.mode == 'demo':
demo.main()
elif FLAGS.mode == 'compile':
train_compile.main(training_folder=trainig_folder)
elif FLAGS.mode == 'taco':
train_taco.main(training_folder=trainig_folder)
else:
logging.fatal('Improper Mode {}'.format(FLAGS.mode))
logging.info('Done')
def score():
#print("请求已开始")
#theimg1 = str(json.loads(request.values.get("theimg1")))
#theimg2=str(json.loads(request.values.get("theimg2")))
theimg1 = str(json.loads(request.values.get("temp1")))
theimg2 = str(json.loads(request.values.get("temp2")))
img1 = base64.b64decode(theimg1)
img2 = base64.b64decode(theimg2)
f1 = open(
'E:/meizhuanghouduan/PSGAN/assets/images/non-makeup/xfsy_0106.png',
'wb')
f1.write(img1)
f1.close()
f2 = open('E:/meizhuanghouduan/PSGAN/assets/images/makeup/makeup.png',
'wb')
f2.write(img2)
f2.close()
demo.main()
f = open('E:/meizhuanghouduan/transferred_image.png', 'rb')
img = base64.b64encode(f.read()).decode('utf-8')
f.close()
#res='返回成功'
print("文件已加载")
return json.dumps(img)
def __init__(self, *args, **kw):
info = main()
info_string = ""
for line in info:
info_string += str(line)
info_string += "\n"
super(HelloFrame, self).__init__(*args, **kw)
pnl = wx.Panel(self)
vbox = wx.BoxSizer(wx.VERTICAL)
st = wx.StaticText(pnl, label=info_string, style=wx.ALIGN_LEFT)
font = wx.Font(10, wx.ROMAN, wx.NORMAL, wx.NORMAL)
st.SetFont(font)
vbox.Add(st, flag=wx.ALL, border=15)
pnl.SetSizer(vbox)
self.Centre()
self.SetSize((800, 800))
self.makeMenuBar()
self.CreateStatusBar()
self.SetStatusText("Welcome to wxPython!")
if __name__ == "__main__":
# solver_init()
cfg_from_list(['CONST.BATCH_SIZE', 1])
for checkpoint_dir, pre_dir, model_name in zip(Checkpoint_dir, Pre_dir,
Model_names):
# print(checkpoint_dir, pre_dir)
weights = os.path.join(checkpoint_dir, 'checkpoint.tar')
set_weights(weights)
set_model_name(model_name)
# if pre_dir == './prediction/ResidualGRUNet_No_Regularition/':
# cfg_from_list(['TEST.VOXEL_THRESH', [0.4]])
for index in range(1, 11):
checkpoint_path = os.path.join(
checkpoint_dir, 'checkpoint.%d.tar' % (index * 2000))
# print(save_path)
symlink_path = os.path.join(checkpoint_dir, 'checkpoint.tar')
if os.path.lexists(symlink_path):
os.remove(symlink_path)
os.symlink("%s" % os.path.abspath(checkpoint_path), symlink_path)
# set_pred_file_name
name = os.path.join(pre_dir, 'prediction.%d.obj' % (index * 2000))
set_pred_file_name(name)
main()
# Make a symlink to the latest network params
# os.symlink("%s" % os.path.abspath(checkpoint_path), symlink_path)
import demo
if __name__ == "__main__":
demo.main()
def script_call():
return redirect(demo.main(app.config['SRC']))
mask_input[np.where(np.logical_and(mask_input == -1,
np.logical_or(result >= 2.5,
target >= 2.5)))] = 0
tracker2 = IndexTracker(ax2, mask_input)
fig.canvas.mpl_connect('scroll_event', tracker2.onscroll)
ax3 = fig.add_subplot(223)
ax3.set_title('Prediction')
tracker3 = IndexTracker(ax3, result)
fig.canvas.mpl_connect('scroll_event', tracker3.onscroll)
if target is not None:
ax4 = fig.add_subplot(224)
ax4.set_title('Target')
tracker4 = IndexTracker(ax4, target)
fig.canvas.mpl_connect('scroll_event', tracker4.onscroll)
plt.tight_layout()
plt.show()
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Slicer')
parser.add_argument('--model', type=str, default='../models/checkpoint.pth',
help='trained model path')
parser.add_argument('--input', type=str, default='../datasets/sample/overfit.h5',
help='uses file as input')
args = parser.parse_args()
main(args.model, args.input, n_samples=1, cb=plot_slicer)
#
#print("back type = ",type(back))
#print("image type = ",type(np.asarray(image)))
#print("masksDL type = ",type(255*mask_sel.astype(np.uint8)))
#
#
#print("back shape = ", back.shape)
#print("image shape = ",np.asarray(image).shape)
#print("masksDL shape = ",255*mask_sel.astype(np.uint8).shape)
#back_align = alignImages(back, np.asarray(image), cv2.cvtColor(255*mask_sel.astype(np.uint8),cv2.COLOR_GRAY2RGB))
#bg_removed = remove_bg(np.asarray(image), back_align,cv2.cvtColor(255*mask_sel.astype(np.uint8),cv2.COLOR_GRAY2RGB))
#config = flags.FLAGS
#config(sys.argv)
# Using pre-trained model, change this to use your own.
#config.load_path = src.config.PRETRAINED_MODEL
#
#config.batch_size = 1
#cv2.imwrite(dir_name.replace('img','back'),remove_bg)
main(bg_removed, args.height, None)
#name= dir_name.replace('img','masksDL')
#cv2.imwrite(name,(255*mask_sel).astype(np.uint8))
#cv2.imwrite(dir_name.replace('img','back'),back_align)
#str_msg='\nDone: ' + dir_name
#print(str_msg)
def test_main():
main()
def modify_doc(doc):
demo.main(doc)
import demo demo.main()
def test_demo(args: List[str]):
demo.main(demo.parser.parse_args(args))
def test_runs():
demo.main()
#
#print("back type = ",type(back))
#print("image type = ",type(np.asarray(image)))
#print("masksDL type = ",type(255*mask_sel.astype(np.uint8)))
#
#
#print("back shape = ", back.shape)
#print("image shape = ",np.asarray(image).shape)
#print("masksDL shape = ",255*mask_sel.astype(np.uint8).shape)
#back_align = alignImages(back, np.asarray(image), cv2.cvtColor(255*mask_sel.astype(np.uint8),cv2.COLOR_GRAY2RGB))
#bg_removed = remove_bg(np.asarray(image), back_align,cv2.cvtColor(255*mask_sel.astype(np.uint8),cv2.COLOR_GRAY2RGB))
#config = flags.FLAGS
#config(sys.argv)
# Using pre-trained model, change this to use your own.
#config.load_path = src.config.PRETRAINED_MODEL
#
#config.batch_size = 1
#cv2.imwrite(dir_name.replace('img','back'),remove_bg)
main(bg_removed, None)
#name= dir_name.replace('img','masksDL')
#cv2.imwrite(name,(255*mask_sel).astype(np.uint8))
#cv2.imwrite(dir_name.replace('img','back'),back_align)
#str_msg='\nDone: ' + dir_name
#print(str_msg)
from demo import main main(['walWalkNext', 'NK', 'rLS', 'fit', '0', '1', '10', '2', '0']) #main(['walWalkNext', 'NK', 'rLS', 'fit', '0', '1', '1000', '2', '0']) #main(['walWalkNext', 'NK', 'rLS', 'fit', '0', '1', '1000', '4', '0']) #main(['walWalkNext', 'NK', 'rLS', 'fit', '0', '1', '1000', '8', '0'])
