def getModel(
config_file
): # No need to specify last_layer as we don't need to do backward pass
with open(config_file, 'r') as f:
config = f.readlines()
Whh = torch.load(config[-5].strip())
Wxh = torch.load(config[-4].strip())
Why = torch.load(config[-3].strip())
Bhh = torch.load(config[-2].strip())
Bhy = torch.load(config[-1].strip())
model = Model()
il = 0
for desc in config[1:-5]:
desc = desc.split()
if desc[0] == 'rnn':
in_features, hidden_features, out_features = int(desc[1]), int(
desc[2]), int(desc[3])
layer = RNN(in_features, hidden_features, out_features)
layer.Whh = torch.Tensor(Whh[il])
layer.Wxh = torch.Tensor(Wxh[il])
layer.Why = torch.Tensor(Why[il])
layer.Bhh = torch.Tensor(Bhh[il]).view(hidden_features, 1)
layer.Bhy = torch.Tensor(Bhy[il]).view(out_features, 1)
il += 1
else:
print(desc[0] + ' layer not implemented!')
model.addLayer(layer)
return model
def analyzeTranslationInvariance():
# setup model
model = Model(open(Constants.fnCharList).read(),
DecoderType.BestPath,
must_restore=True)
# read image and specify ground-truth text
img = cv2.imread(Constants.fnAnalyze, cv2.IMREAD_GRAYSCALE)
(w, h) = img.shape
assert Model.imgSize[1] == w
imgList = []
for dy in range(Model.imgSize[0] - h + 1):
targetImg = np.ones((Model.imgSize[1], Model.imgSize[0])) * 255
targetImg[:, dy:h + dy] = img
imgList.append(preprocess(targetImg, Model.imgSize))
# put images and gt texts into batch
batch = Batch([Constants.gtText] * len(imgList), imgList)
# compute probabilities
(texts, probs) = model.infer_batch(batch,
calc_probability=True,
probability_of_gt=True)
# save results to file
f = open(Constants.fnTranslationInvarianceTexts, 'wb')
pickle.dump(texts, f)
f.close()
np.save(Constants.fnTranslationInvariance, probs)
def getModel(ident):
if ident == "SMD":
print("Retrieved: spring-mass-damper system" +\
"\n p[0] k/m" +\
"\n p[1] d/m"+\
"\n p[2] 1/m\n")
M = Model("Spring-Mass-Damper System",'ltiss',dt,[2,1,1,3],f_smd, A_smd, B_smd,\
F_smd, h_smd, C_smd, D_smd, H_smd)
return M
elif ident == "NLSMD":
print("Retrieved: non-linear spring-mass-damper system" +\
"\n p[0] k/m" +\
"\n p[1] d/m"+\
"\n p[2] 1/m\n")
M = Model("Non-linear Spring-Mass-Damper System",dt,[2,1,1,3],f_nlsmd, \
dfx_nlsmd, dfp_nlsmd, h_nlsmd, dhx_nlsmd, dhp_nlsmd)
return M
elif ident == "PEN":
print("Retrieved: pendulum system" +\
"\n p[0] g/l" +\
"\n p[1] d/(ml^2)"+\
"\n p[2] 1/(ml^2)\n")
M = Model("Pendulum without spring",dt,[2,1,1,3],f_pen, \
dfx_pen, dfp_pen, h_pen, dhx_pen, dhp_pen)
return M
else:
print("Unknown system identifier. Options are: "+\
"\n SMD spring-mass-damper system" +\
"\n NLSMD Non-linear spring-mass-damper system" +\
"\n PEN Pendulum\n" )
def getModel(config_file):
with open(config_file,'r') as f:
config = f.readlines()
weights = torchfile.load(config[-2].strip())
biases = torchfile.load(config[-1].strip())
model = Model()
il = 0
for desc in config[1:-2]:
desc = desc.split()
if desc[0] == 'linear':
in_features, out_features = int(desc[1]), int(desc[2])
layer = Linear(in_features, out_features)
layer.W = torch.Tensor(weights[il])
layer.B = torch.Tensor(biases[il]).view(out_features, 1)
il += 1
elif desc[0] == 'relu':
layer = ReLU()
elif desc[0] == 'dropout':
layer = Dropout(float(desc[1]), isTrain=False)
else:
print(desc[0] + ' layer not implemented!')
model.addLayer(layer)
return model
def test_normal_use(self):
m = Model(3500, 100000)
m.stepUp()
self.assertEqual(m.getFrequency(), 3600)
m.stepDown()
self.assertEqual(m.getFrequency(), 3500)
self.assertEqual(m.getFrequency(FrequencyUnit.HZ), 3500000)
self.assertEqual(m.getFrequency(FrequencyUnit.MHZ), 3.5)
def __init__(self):
self.inpt= { # just the default to start with
"wind": 3,
"solar": 500,
"albedo": 0.25,
"airt": 15,
"sfc": "grass (dry)",
"rs": 40,
"vp": 10,
"smd": 10
}
self.rblist = [500.0, -30.0, 300.0, 200.0]
self.eblist = [300.0, 100.0, 150.0, 200.0]
# Polynomial fit to Graham Russell's smd fit for cereals
# yfit = a*x**3+b*x+c where x is smd in mm
self.smdfit = [1.27791987e-04, -9.56796959e-02, 3.95338027e+01]
self.mod = Model(self.inpt)
self.tlist = self.mod.tlist
self.vw = View(self.rblist, self.eblist, self.tlist)
self.sfcs = ["grass (dry)","bare soil (dry)","cereals (dry)",
"conifers (dry)","upland (dry)","grass (wet)", "bare soil (wet)",
"cereals (wet)","conifers (wet)", "upland (wet)", "water"]
self.bit_wind = widgets.BoundedIntText(value = self.inpt["wind"], min=1, max=15, step=1,
description="u ($m \ s^{-1}$)", width=50)
self.bit_solar = widgets.BoundedIntText(value = self.inpt["solar"], min=1, max=1000, step=10,
description="solar ($W m^{-2}$)", width=50)
self.bit_vp = widgets.BoundedIntText(value =self.inpt["vp"], min=1, max=40, step=1,
description="vp (mbar)", width=50)
self.dd_surface = widgets.Dropdown(value =self.inpt["sfc"], options=self.sfcs,
description="surface", width=50)
self.bit_smd = widgets.BoundedIntText(value=self.inpt["smd"], min=1, max=180, step=5,
description="smd (mm)", width=50)
self.bit_airt = widgets.BoundedIntText(value=self.inpt["airt"], min=-5, max=40, step=1,
description="air T (oC)", width=50)
self.txt_rs = widgets.Text(description="rs")
self.txt_rh = widgets.Text(description="RH (%)")
self.txt_le = widgets.Text(description="LE")
self.txt_ra = widgets.Text(description="ra")
# First time round to populate output boxes
self.rblist, self.eblist, self.tlist, self.olist = self.mod.calculateLE(self.inpt)
self.txt_rs.value = str('{0:.0f}'.format(self.olist[0]))
self.txt_rh.value = str('{0:.1f}'.format(self.olist[1]))
self.txt_le.value = str('{0:.1f}'.format(self.olist[2]))
self.txt_ra.value = str('{0:.0f}'.format(self.olist[3]))
self.bit_wind.observe(self.bit_wind_eventhandler, names='value')
self.bit_solar.observe(self.bit_solar_eventhandler, names='value')
self.bit_vp.observe(self.bit_vp_eventhandler, names='value')
self.dd_surface.observe(self.dd_surface_eventhandler, names='value')
self.bit_smd.observe(self.bit_smd_eventhandler, names='value')
self.bit_airt.observe(self.bit_airt_eventhandler, names='value')
self.h0 = widgets.HBox(children=[self.dd_surface, self.bit_smd])
self.h1 = widgets.HBox(children=[self.bit_solar, self.bit_wind])
self.h2 = widgets.HBox(children=[self.bit_airt, self.bit_vp, self.txt_rh])
self.h3 = widgets.HBox(children=[self.txt_ra, self.txt_rs, self.txt_le])
def main():
"main function"
# optional command line args
parser = argparse.ArgumentParser()
parser.add_argument('--train', help='train the NN', action='store_true')
parser.add_argument('--validate',
help='validate the NN',
action='store_true')
parser.add_argument('--beamsearch',
help='use beam search instead of best path decoding',
action='store_true')
parser.add_argument(
'--wordbeamsearch',
help='use word beam search instead of best path decoding',
action='store_true')
parser.add_argument('--dump',
help='dump output of NN to CSV file(s)',
action='store_true')
args = parser.parse_args()
decoderType = DecoderType.BestPath
if args.beamsearch:
decoderType = DecoderType.BeamSearch
elif args.wordbeamsearch:
decoderType = DecoderType.WordBeamSearch
# train or validate on IAM dataset
if args.train or args.validate:
# load training data, create TF model
loader = DataLoader(FilePaths.fnTrain, Model.batchSize, Model.imgSize,
Model.maxTextLen)
# save characters of model for inference mode
open(FilePaths.fnCharList, 'w').write(str().join(loader.charList))
# save words contained in dataset into file
open(FilePaths.fnCorpus, 'w').write(
str(' ').join(loader.trainWords + loader.validationWords))
# execute training or validation
if args.train:
model = Model(loader.charList, decoderType)
train(model, loader)
elif args.validate:
model = Model(loader.charList, decoderType, mustRestore=True)
validate(model, loader)
# infer text on test image
else:
print(open(FilePaths.fnAccuracy).read())
model = Model(open(FilePaths.fnCharList).read(),
decoderType,
mustRestore=True,
dump=args.dump)
infer(model, FilePaths.fnInfer)
def __init__(self, config, logger, metadata):
super(Task, self).__init__(config, logger, metadata)
self.sum = 0
self.model = Model(config.get_with_prefix("model"))
self.data = Data(config.get_with_prefix("data"))
self.trainer = Trainer(config.get_with_prefix("trainer"), self.model,
self.data)
self.best_val_acc = 0
self.number_worse_iterations = 0
def plot_model(contact_matrix,
susceptible,
infectious_rate,
days,
reference_cases=None,
measure_factor: float = 1,
measure_day=0,
reference_hospital=None,
offset=0,
first_patient_age=38,
name='model',
scenario=None,
cap_ic=False,
title='') -> Model:
model = Model(contact_matrix,
susceptible,
infectious_rate,
measure_factor=measure_factor,
measure_day=measure_day,
first_patient_age=first_patient_age,
scenario=scenario,
cap_ic=cap_ic)
model.run(days)
y = [
model.infected_data, model.recovered_data, model.exposed_data,
model.hospital_data, model.ic_data, model.dead_data
]
labels = ['infected', 'recovered', 'exposed', 'hospitalized', 'ic', 'dead']
plot(y, labels, name, 'Amount', title=title)
if reference_cases is not None:
y = [reference_cases, model.case_data]
labels = ['Reference Cases', 'Model Cases']
plot(y,
labels,
'ref_cases',
'Amount',
title='Verschil model werkelijkheid')
if reference_hospital is not None:
y = [
np.append(np.zeros(offset), reference_hospital),
model.hospital_data + model.ic_data
]
labels = ['Reference Hospital + ic', 'Model Hospital + ic']
plot(y,
labels,
'ref_hospital',
'Amount',
title='Verschil model werkelijkheid')
return model
def analyzePixelRelevance():
"simplified implementation of paper: Zintgraf et al - Visualizing Deep Neural Network Decisions: Prediction Difference Analysis"
# setup model
model = Model(open(Constants.fnCharList).read(), DecoderType.BestPath, mustRestore=True)
# read image and specify ground-truth text
img = cv2.imread(Constants.fnAnalyze, cv2.IMREAD_GRAYSCALE)
(w, h) = img.shape
assert Model.imgSize[1] == w
# compute probability of gt text in original image
batch = Batch([Constants.gtText], [preprocess(img, Model.imgSize)])
(_, probs) = model.inferBatch(batch, calcProbability=True, probabilityOfGT=True)
origProb = probs[0]
grayValues = [0, 63, 127, 191, 255]
if Constants.distribution == 'histogram':
bins = [0, 31, 95, 159, 223, 255]
(hist, _) = np.histogram(img, bins=bins)
pixelProb = hist / sum(hist)
elif Constants.distribution == 'uniform':
pixelProb = [1.0 / len(grayValues) for _ in grayValues]
else:
raise Exception('unknown value for Constants.distribution')
# iterate over all pixels in image
pixelRelevance = np.zeros(img.shape, np.float32)
for x in range(w):
for y in range(h):
# try a subset of possible grayvalues of pixel (x,y)
imgsMarginalized = []
for g in grayValues:
imgChanged = copy.deepcopy(img)
imgChanged[x, y] = g
imgsMarginalized.append(preprocess(imgChanged, Model.imgSize))
# put them all into one batch
batch = Batch([Constants.gtText]*len(imgsMarginalized), imgsMarginalized)
# compute probabilities
(_, probs) = model.inferBatch(batch, calcProbability=True, probabilityOfGT=True)
# marginalize over pixel value (assume uniform distribution)
margProb = sum([probs[i] * pixelProb[i] for i in range(len(grayValues))])
pixelRelevance[x, y] = weightOfEvidence(origProb, margProb)
print(x, y, pixelRelevance[x, y], origProb, margProb)
np.save(Constants.fnPixelRelevance, pixelRelevance)
def mainView(self):
students = FileBoundStudentsFactory().create()
students.retrieveState()
model = Model(students)
mainView = MainView(self.root, model)
mainView.pack()
self.root.mainloop()
def createModel(spec_file):
with open(spec_file, 'r') as f:
spec = f.readlines()
lenspec = len(spec)
model = Model()
num_layers = 0
for desc_ind in range(lenspec):
desc = spec[desc_ind]
desc = desc.split()
if desc[0] == 'rnn':
in_features, hidden_features, out_features = int(desc[1]), int(
desc[2]), int(desc[3])
layer = RNN(in_features, hidden_features, out_features)
num_layers += 1
else:
print(desc[0] + ' layer not implemented!')
model.addLayer(layer)
return model, (spec, num_layers)
def before_first_request():
global model, graph
graph = tf.Graph()
weight_location = 'weights'
with graph.as_default():
model = Model(vocabulary,
graph,
DecoderType.BestPath,
model_dir=weight_location)
def createModel(spec_file):
with open(spec_file,'r') as f:
spec = f.readlines()
model = Model()
num_linear_layers = 0
for desc in spec:
desc = desc.split()
if desc[0] == 'linear':
in_features, out_features = int(desc[1]), int(desc[2])
layer = Linear(in_features, out_features)
num_linear_layers += 1
elif desc[0] == 'relu':
layer = ReLU()
elif desc[0] == 'dropout':
layer = Dropout(float(desc[1]), isTrain=True)
else:
print(desc[0] + ' layer not implemented!')
model.addLayer(layer)
return model, (spec, num_linear_layers)
def create_model(config_file):
model = Model()
with open(config_file, 'r') as f:
num_layers = int(f.readline().strip())
for i in range(num_layers):
layer_info = f.readline().strip().split(' ')
layer_type = layer_info[0]
if layer_type == LINEAR:
num_inputs = int(layer_info[1])
num_outputs = int(layer_info[2])
model.addLayer(Linear(num_inputs, num_outputs))
elif layer_type == RELU:
model.addLayer(ReLU())
weight_file = f.readline().strip()
bias_file = f.readline().strip()
weights = load_file(weight_file)
biases = load_file(bias_file)
linear_index = 0
for layer in model.Layers:
if isinstance(layer, Linear):
layer.W = weights[linear_index]
layer.B = biases[linear_index]
linear_index += 1
return model
def test_update(self):
m = Model(3500, 100)
do = DefaultOutput()
m.registerOutput(do)
m.start()
self.assertEqual(3500, do.lastFreq)
def create_model(self):
print("Create model called")
if self.model is None:
self.model = Model(self, None,
self.importExportDataManager.get_data())
if self.getSelectedAlgorithm() == "Logistic Regression":
self.model.set_model(
LogisticRegression(solver='lbfgs', max_iter=1000))
elif self.getSelectedAlgorithm() == "Decision Tree":
self.model.model = DecisionTreeClassifier()
print(self.model.model.__class__)
elif self.getSelectedAlgorithm() == "Naive Bayes":
self.model.set_model(GaussianNB())
else:
messagebox.showinfo("Warning", "No algorithm selected")
return
optimizeParamsMsgBox = messagebox.askquestion(
"Optimize hyperparameters",
"Would you like PyMine to optimize the parameters for this model?")
if optimizeParamsMsgBox == 'yes':
self.model.optimize_model_hyperparams(self.model.model)
self.model.performance_summary()
def main():
""" Main function """
# Opptional command line args
parser = argparse.ArgumentParser()
parser.add_argument(
"--train", help="train the neural network", action="store_true")
parser.add_argument(
"--validate", help="validate the neural network", action="store_true")
parser.add_argument(
"--wordbeamsearch", help="use word beam search instead of best path decoding", action="store_true")
args = parser.parse_args()
decoderType = DecoderType.BestPath
if args.wordbeamsearch:
decoderType = DecoderType.WordBeamSearch
# Train or validate on Cinnamon dataset
if args.train or args.validate:
# Load training data, create TF model
loader = DataLoader(FilePaths.fnTrain, Model.batchSize,
Model.imgSize, Model.maxTextLen, load_aug=True)
# Execute training or validation
if args.train:
model = Model(loader.charList, decoderType)
train(model, loader)
elif args.validate:
model = Model(loader.charList, decoderType, mustRestore=False)
validate(model, loader)
# Infer text on test image
else:
print(open(FilePaths.fnAccuracy).read())
model = Model(open(FilePaths.fnCharList).read(),
decoderType, mustRestore=False)
infer(model, FilePaths.fnInfer)
def __init__(self):
""" Controller Class constructor """
# Cmd Line Argument Setup
self.arg_parser = ArgParserFactory.create_arg_parser()
self.args = self.arg_parser.parse_args()
# Config File Setup
self.config_handler = ConfigHandler()
self.config_settings = self.config_handler.read_config(
'../config/config.json')
# File Factory
self.file_factory = FileFactory
self.directory_factory = DirectoryFactory
# Empty Model
self.model = Model()
def infer_by_web(path, option):
decoderType = DecoderType.BestPath
print(open(FilePaths.fnAccuracy).read())
model = Model(open(FilePaths.fnCharList).read(),
decoderType, mustRestore=False)
# show_wait_destroy2("received", path)
print("Passing image to read for handwritten : ", path); # modified 08092020
recognized = infer(model, path)
# morph = path.copy()
# show_wait_destroy2("morph_received", morph)
# print("Passing image to read for handwritten : ", morph); # modified 08092020
# recognized = infer(model, morph)
return recognized
def infer_samples(decoder_type=DecoderType.BestPath):
# print(open(FilePaths.fnAccuracy).read())
model = Model(open(FilePaths.fnCharList).read(),
decoder_type,
must_restore=True)
path = os.path.join('data', 'test_images')
image_files = [os.path.join(path, f) for f in os.listdir(path)]
# print(image_files)
print("######### Inferring Text from Test Images###############")
for img in image_files:
# print(img)
try:
infer(model, img, show_img=2)
except Exception as e:
print("Exception: " + str(e))
pass
def normalRun(step,freq,input,output,verbose):
model = Model(freq, step)
inputMethod, outputMethod = None, None
if input=='cli':
inputMethod = CliInput(model)
if output=='cli':
outputMethod = CliOutput()
model.registerOutput(outputMethod)
model.start()
inputMethod.start()
class Task(taskplan.Task):
def __init__(self, config, logger, metadata):
super(Task, self).__init__(config, logger, metadata)
self.sum = 0
self.model = Model(config.get_with_prefix("model"))
self.data = Data(config.get_with_prefix("data"))
self.trainer = Trainer(config.get_with_prefix("trainer"), self.model,
self.data)
self.best_val_acc = 0
self.number_worse_iterations = 0
def save(self, path):
self.model.save_weights(str(path / "model.h5py"))
pickle.dump(self.best_val_acc, open(str(path / "best_model.pkl"),
"wb"))
def step(self, tensorboard_writer, current_iteration):
with tensorboard_writer.as_default():
val_acc = self.trainer.step(current_iteration)
if val_acc is not None:
if val_acc > self.best_val_acc:
self.best_val_acc = val_acc
self.model.save_weights(str(self.task_dir / "model.h5py"))
self.number_worse_iterations = 0
else:
self.number_worse_iterations += 1
if self.number_worse_iterations > 5:
self.pause_computation = True
tf.summary.scalar('val/best_acc',
self.best_val_acc,
step=current_iteration)
def load(self, path):
self.model.load_weights(str(path / "model.h5py"))
self.best_val_acc = pickle.load(
open(str(path / "best_model.pkl"), "rb"))
def generalize(M,p,S=[]):
# Fetch everything from model
f = M.f;
A = M.A;
B = M.B;
F = M.F;
h = M.h;
C = M.C;
D = M.D;
H = M.H;
th = M.th;
nx = M.nx
nu = M.nu;
ny = M.ny;
nth = M.nth;
name = M.name;
typ = M.typ;
dt = M.dt;
Q = M.Q;
R = M.R;
# Generalize all system functions and matrices
# State transition function
def f_t(x,u,th):
x_t = zeros(p*nx);
for i in range(0,p):
x_t[i*nx:(i+1)*nx] = f(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return x_t;
# State transition function to state gradient
def A_t(x,u,th):
A_t = zeros((p*nx,p*nx));
for i in range(0,p):
A_t[i*nx:(i+1)*nx,i*nx:(i+1)*nx] = \
A(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return A_t;
# State transition function to input gradient
def B_t(x,u,th):
B_t = zeros((p*nx,p*nu));
for i in range(0,p):
B_t[i*nx:(i+1)*nx,i*nu:(i+1)*nu] = \
B(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return B_t;
# State transition function to parameter gradient
def F_t(x,u,th):
F_t = zeros((p*nx,nth));
for i in range(0,p):
F_t[i*nx:(i+1)*nx,:] = \
F(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return F_t;
# Output function
def h_t(x,u,th):
y_t = zeros(p*ny);
for i in range(0,p):
y_t[i*nx:(i+1)*nx] = h(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return y_t;
# Output function to state gradient
def C_t(x,u,th):
C_t = zeros((p*ny,p*nx));
for i in range(0,p):
C_t[i*ny:(i+1)*ny,i*nx:(i+1)*nx] = \
C(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return C_t;
# Output function to input gradient
def D_t(x,u,th):
D_t = zeros((p*ny,p*nu));
for i in range(0,p):
D_t[i*ny:(i+1)*ny,i*nu:(i+1)*nu] = \
D(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return D_t;
# Measurement function to parameter gradient
def H_t(x,u,th):
H_t = zeros((p*ny,nth));
for i in range(0,p):
H_t[i*ny:(i+1)*ny,:] = \
H(x[i*nx:(i+1)*nx],u[i*nu:(i+1)*nu],th);
return H_t;
# Include corellation information if specified
if not isempty(Q) and not isempty(R):
if not isempty(S):
Pw = kron(S,inv(Q));
Pz = kron(S,inv(R));
else:
Pw = kron(eye(p),inv(Q));
Pz = kron(eye(p),inv(R));
else:
Pw = [];
Pz = [];
M_t = Model('Generalized '+ name, typ, dt, [p*nx,p*nu,p*ny,nth], \
f_t,A_t,B_t,F_t,h_t,C_t,D_t,H_t,th,Pw,Pz);
return M_t, kron(eye(p,p,1),eye(int(M_t.nx/p)));
def main():
print("⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆")
print("★ Welcome to Language Detection by No Look Pass ★")
print("⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆⋆")
training_set = "OriginalDataSet/training-tweets.txt"
# Training Set is fixed for the purposes of the demo
# while True:
# print("Enter Training Set file path: ", end="")
# training_set = input()
# training_set = "OriginalDataSet/training-tweets.txt"
# # Source: https://www.guru99.com/python-check-if-file-exists.html
# if path.exists(training_set):
# print("\tOK: Training File Accepted\n")
# break
# else:
# print("\tERROR: File path does not exist\n")
while True:
print("Enter Test Set file path: ", end="")
test_set = input()
# test_set = "OriginalDataSet/test-tweets-given.txt"
# Source: https://www.guru99.com/python-check-if-file-exists.html
if path.exists(test_set):
print("\tOK: Test File Accepted\n")
break
else:
print("\tERROR: File path does not exist\n")
# Create output folder
create_directory("output/")
# Initialize models
byom = Model(v=3, n=3, delta=0.1)
model_v0_n1_d0 = Model(v=0, n=1, delta=0)
model_v1_n2_d05 = Model(v=1, n=2, delta=0.5)
model_v1_n3_d1 = Model(v=1, n=3, delta=1)
model_v2_n2_d03 = Model(v=2, n=2, delta=0.3)
# Train models
model_v0_n1_d0.train(training_set)
model_v1_n2_d05.train(training_set)
model_v1_n3_d1.train(training_set)
model_v2_n2_d03.train(training_set)
byom.train(training_set)
# Test models using test set
model_v0_n1_d0.test(test_set)
model_v1_n2_d05.test(test_set)
model_v1_n3_d1.test(test_set)
model_v2_n2_d03.test(test_set)
byom.test(test_set)
# Evaluate models
model_v0_n1_d0.evaluate()
model_v1_n2_d05.evaluate()
model_v1_n3_d1.evaluate()
model_v2_n2_d03.evaluate()
byom.evaluate()
def infer_model(char_list, infer_location):
model = Model(char_list, DecoderType.BestPath)
infer(model, infer_location)
rec=recognized[0], proba=probability[0]))
def train_model(model, train_dir):
loader = DataLoader(Model.batchSize, Model.imgSize, Model.maxTextLen,
train_dir)
logging.info('Model chars: {chars}.'.format(chars=loader.charList))
train(model, loader)
def infer_model(char_list, infer_location):
model = Model(char_list, DecoderType.BestPath)
infer(model, infer_location)
if __name__ == '__main__':
logging.basicConfig(level=logging.INFO,
format='%(asctime)s;%(levelname)s;%(message)s')
vocabulary = '.0123456789'
in_dir = 'C:/Users/FlorijnWim/PycharmProjects/htr-ctctcnn/traindata'
output_model_dir = 'C:/Users/FlorijnWim/PycharmProjects/htr-ctctcnn/model_new'
graph = tf.Graph()
with graph.as_default():
output_model = Model(vocabulary,
graph,
DecoderType.BestPath,
model_dir=output_model_dir)
train_model(output_model, in_dir)
class Mainframe(Frame):
"""
Multiple Docuement Interface
Everything in the application stems from the mainframe
"""
def __init__(self, root):
self.root = root
self.importExportDataManager = ImportExportDataManager(self)
self.dataPreprocessingManager = DataPreprocessingManager(self)
self.discretizationManager = DiscretizationManager(self)
self.optionsWindow = OptionsWindow(self)
self.frame = Frame(root)
self.model = None
# create menu bar
self.menubar = Menu(root)
root.config(menu=self.menubar)
self.fileMenu = Menu(self.menubar, tearoff=False)
self.menubar.add_cascade(label="File", menu=self.fileMenu)
self.fileMenu.add_command(
label="Import data...",
command=self.importExportDataManager.set_filename)
self.fileMenu.add_command(label="Options...",
command=self.optionsWindow.create)
self.dataMenu = Menu(self.menubar, tearoff=False)
self.menubar.add_cascade(label="Data", menu=self.dataMenu)
self.dataMenu.add_command(label="Replace missing values...",
command=self.dataPreprocessingManager.create)
self.dataMenu.add_command(label="Discretize data...",
command=self.discretizationManager.create)
# create labels
self.lblReduceFeatures = Label(self.frame, text="Reduce Features?")
self.lblInterpolateMissingValues = Label(
self.frame, text="Interpolate missing values?")
self.lblAttributes = Label(self.frame, text="Select class label: ")
self.lblAlgorithm = Label(self.frame, text="Select algorithm: ")
# create comboboxes
self.cmbReduceFeatures = Combobox(self.frame,
state="readonly",
values=["Yes", "No"],
width=5)
self.cmbInterpolateMissingValues = Combobox(self.frame,
state="readonly",
values=["Yes", "No"],
width=5)
self.cmbAttributes = Combobox(self.frame, state="readonly")
self.cmbAlgorithm = Combobox(
self.frame,
state="readonly",
values=["Logistic Regression", "Decision Tree", "Naive Bayes"])
# set default combobox values
self.cmbReduceFeatures.set("No")
self.cmbInterpolateMissingValues.set("No")
self.cmbAlgorithm.set("None")
# create preview data table
self.previewDataTable = PreviewDataTable(self)
# create buttons
self.runButton = Button(text="Run", width=10)
self.runButton.bind("<Button-1>", self.run_algorithm)
# set grid layout
rowNumber = 0
self.frame.grid(row=rowNumber, column=1, sticky=W)
self.lblReduceFeatures.grid(row=rowNumber, column=1, sticky=W)
self.cmbReduceFeatures.grid(row=rowNumber, column=2, sticky=W)
rowNumber += 1
self.lblInterpolateMissingValues.grid(row=rowNumber,
column=1,
sticky=W)
self.cmbInterpolateMissingValues.grid(row=rowNumber,
column=2,
sticky=W)
rowNumber += 1
self.lblAttributes.grid(row=rowNumber, column=1, sticky=W)
self.cmbAttributes.grid(row=rowNumber, column=2, sticky=W)
rowNumber += 1
self.lblAlgorithm.grid(row=rowNumber, column=1, sticky=W)
self.cmbAlgorithm.grid(row=rowNumber, column=2, sticky=W)
rowNumber += 1
Label(root, text="").grid(row=rowNumber, column=1)
rowNumber += 1
self.previewDataTable.treeview.grid(
row=rowNumber,
column=1,
rowspan=self.previewDataTable.treeview.winfo_width(),
columnspan=2,
sticky=W)
rowNumber += 1
Label(root, text="").grid(row=rowNumber, column=1)
rowNumber += 1
self.runButton.grid(row=rowNumber, column=1)
rowNumber += 1
def getReduceFeatureOption(self):
return self.cmbReduceFeatures.get()
def getInterpolateMissingValuesOption(self):
return self.cmbInterpolateMissingValues.get()
def getSelectedClassLabel(self):
return self.cmbAttributes.get()
def getSelectedAlgorithm(self):
return self.cmbAlgorithm.get()
def getSelectedUserParams(self, event):
print("Reduce features: " + self.getReduceFeatureOption())
print("Interpolate missing values: " +
self.getInterpolateMissingValuesOption())
print("Class label: " + self.getSelectedClassLabel())
print("Selected algorithm: " + self.getSelectedAlgorithm())
print(self.importExportDataManager.summary())
def run_algorithm(self, event):
if self.importExportDataManager.get_filename() is not None and \
isinstance(self.importExportDataManager.get_data(), pd.DataFrame):
if self.getReduceFeatureOption() == 'Yes':
print("Reducing features...")
if self.getInterpolateMissingValuesOption() == 'Yes':
print("Interpolating values...")
self.create_model()
def create_model(self):
print("Create model called")
if self.model is None:
self.model = Model(self, None,
self.importExportDataManager.get_data())
if self.getSelectedAlgorithm() == "Logistic Regression":
self.model.set_model(
LogisticRegression(solver='lbfgs', max_iter=1000))
elif self.getSelectedAlgorithm() == "Decision Tree":
self.model.model = DecisionTreeClassifier()
print(self.model.model.__class__)
elif self.getSelectedAlgorithm() == "Naive Bayes":
self.model.set_model(GaussianNB())
else:
messagebox.showinfo("Warning", "No algorithm selected")
return
optimizeParamsMsgBox = messagebox.askquestion(
"Optimize hyperparameters",
"Would you like PyMine to optimize the parameters for this model?")
if optimizeParamsMsgBox == 'yes':
self.model.optimize_model_hyperparams(self.model.model)
self.model.performance_summary()
def run(filename):
"main function"
# optional command line args
parser = argparse.ArgumentParser()
parser.add_argument('--train', help='train the NN', action='store_true')
parser.add_argument('--validate',
help='validate the NN',
action='store_true')
parser.add_argument('--beamsearch',
help='use beam search instead of best path decoding',
action='store_true')
parser.add_argument(
'--wordbeamsearch',
help='use word beam search instead of best path decoding',
action='store_true')
parser.add_argument('--dump',
help='dump output of NN to CSV file(s)',
action='store_true')
args = parser.parse_args()
decoderType = DecoderType.BestPath
if args.beamsearch:
decoderType = DecoderType.BeamSearch
elif args.wordbeamsearch:
decoderType = DecoderType.WordBeamSearch
# train or validate on IAM dataset
if args.train or args.validate:
# load training data, create TF model
loader = DataLoader(FilePaths.fnTrain, Model.batchSize, Model.imgSize,
Model.maxTextLen)
# save characters of model for inference mode
open(FilePaths.fnCharList, 'w').write(str().join(loader.charList))
# save words contained in dataset into file
open(FilePaths.fnCorpus, 'w').write(
str(' ').join(loader.trainWords + loader.validationWords))
# execute training or validation
if args.train:
model = Model(loader.charList, decoderType)
train(model, loader)
elif args.validate:
model = Model(loader.charList, decoderType, mustRestore=True)
validate(model, loader)
# infer text on test image
else:
index_list = []
result_list = []
prob_list = []
print(open(FilePaths.fnAccuracy).read())
model = Model(open(FilePaths.fnCharList).read(),
decoderType,
mustRestore=True,
dump=args.dump)
for dirpath, dirnames, files in os.walk('../output_words/' + filename,
topdown=False):
for sub_file in sorted(files, key=getint):
img_path = dirpath + '/' + sub_file
# print('---------------------------------------------------')
index_number, _ = str(sub_file).split('.')
# print("File path: "+img_path)
try:
result, prob = infer(model, img_path)
except ValueError:
print("Value error")
continue
# print(index_number, result, prob)
index_list.append(index_number)
result_list.append(result)
prob_list.append(prob)
return index_list, result_list, prob_list
with open('settings.yml', 'r') as f:
SETTINGS = yaml.load(f, Loader=yaml.FullLoader)
# Start up interface
gui = Interface(SETTINGS)
gui.startup_window()
# After pressing start in startup_window load in whole program
from src.Loader import Loader
from src.Model import Model
# Initialize reader loading new images
reader = Loader(SETTINGS)
# Define model and load in weights
predictor = Model(SETTINGS)
nn = predictor.load()
test_data_loaded = False #indication of the loaded dataset
gui.startup_window.close()
window = gui.window
# Event loop. Read buttons, make callbacks
while True:
# Read the Window
event, value = window.read()
if event in ('Quit', None):
break
# Lookup event in function dictionary
elif event in gui.dispatch_dictionary:
