def index(request):
label = 'Not Start'
global accel_data
dtw = DTW()
dtw.tamplate_load('model.npy')
threshold = 20
if (request.method == 'POST'):
received_json_data = json.loads(request.body)
accel_data.append(
Vector(
(received_json_data['accel_x'], received_json_data['accel_y'],
received_json_data['accel_z'])))
if (len(accel_data) == 150):
#amplitudes = list(map(lambda v: v.modulus(), accel_data))
#print(amplitudes)
#if (max(amplitudes) - min(amplitudes) >= threshold):
label = dtw.test(accel_data, cosine_cost)
accel_data = []
return HttpResponse(label)
else:
'''
motion_period = 5 # 5s
frequency = 50 # 50Hz
period = motion_period * frequency
gesture_dataflow_path = 'simulation.csv'
gestures = pre.data_segmentation(gesture_dataflow_path, period, 2, 5)
for i in range(0, len(gestures)):
label = dtw.test(gestures[i], cosine_cost)
'''
return HttpResponse('Wrong POST')
def get_dtw_alignment(feat1, feat2):
distance_array = cosine_distance(feat1, feat2)
_, _, paths = DTW(feat1, feat2, return_alignment=True,
dist_array=distance_array)
path1, path2 = paths[1:]
assert len(path1) == len(path2)
return path1, path2
def score_ntdw():
''' Calculate ntdw between different experimental settings for paper Table 3. '''
EXPS = [
'sim_results/baseline/submit_coda.json',
'sim_results/baseline/submit_coda_theta_1.json',
'sim_results/baseline/submit_coda_theta_2.json',
'sim_results/baseline/submit_coda_theta_3.json',
'sim_results/baseline_color_jittered/submit_coda_theta_1.json',
'sim_results/baseline_color_jittered/submit_coda_theta_2.json',
'sim_results/baseline_color_jittered/submit_coda_theta_3.json',
'bags/wmap/submit_coda_robot_wmap.json', # Robot with map
'bags/nomap/submit_coda_robot_nomap.json'
] # Robot no map
ntdw = np.zeros((len(EXPS), len(EXPS)))
scorer = DTW()
print('\nCalculating ntdw matrix for %s' % EXPS)
for i, ref_path in enumerate(EXPS):
with open('src/vln_evaluation/data/%s' % ref_path) as f:
ref = json.load(f)
for item in ref:
item['path_id'] = int(item['instr_id'].split('_')[0])
evaluator = SimEvaluation(ref, CONN)
for j, pred_path in enumerate(EXPS):
with open('src/vln_evaluation/data/%s' % pred_path) as f:
pred = json.load(f)
summary, scores = evaluator.score(pred)
ntdw[i, j] = summary['ndtw']
print(ntdw)
def half_asym_fn(A, B):
return DTW.dist_damn_adv(
A,
B,
[(-1, -3), (-1, -2), (-1, -1), (-2, -1), (-3, -1)],
[[1 / 3, 1 / 3, 1 / 3], [1 / 2, 1 / 2], [1], [1, 1], [1, 1, 1]],
)
def __init__(self, gt_data, nav_graph_dir):
self.error_margin = 3.0
self.dtw = DTW(threshold=self.error_margin)
self.graphs = load_nav_graph(nav_graph_dir)
self.gt = {}
self.instr_ids = []
for item in gt_data:
if 'scan' not in item:
item['scan'] = 'yZVvKaJZghh'
if 'trajectory' in item:
item['trajectory'] = self.path_to_points(
item['trajectory'], item['scan'])
if 'path' in item:
item['trajectory'] = self.path_to_points(
item['path'], item['scan'])
if 'instr_id' in item:
self.gt[item['instr_id']] = item
else:
for i in range(3):
self.gt['%s_%d' % (item['path_id'], i)] = item
self.instr_ids += [
'%d_%d' % (item['path_id'], i) for i in range(3)
]
self.instr_ids = set(self.instr_ids)
self.distances = {}
# compute all shortest paths
self.distances = {}
for scan, G in self.graphs.items(): # compute all shortest paths
self.distances[scan] = dict(nx.all_pairs_dijkstra_path_length(G))
def calculate_signature_dissimilarity(self, target_signature):
"""
Calculate the dissimilarity of target_signature the the user's enrolment signatures
:param target_signature: list of the normalized feature vectors of the target signature to compare to this
user's enrolment signatures
:return: dictionary of type
<user enrolment signature id> -> <dissimilarity of target signature to that enrolment signature>
"""
dtw = DTW()
dissimilarities = []
for key, enrolment_signature in self.enrolment_signatures.items():
dissim, matrix = dtw.distance(enrolment_signature,
target_signature, self.WINDOW_SIZE)
dissimilarities.append(dissim)
return np.mean(np.asarray(dissimilarities))
def RunDTW(self):
matrizDistancia = []
vetorResultado = []
# Faz a leitura da base de dados e vai comparando
for linha in self.arquivo:
vetorResultado = [0,0,0]
valoresLinha = linha.split(" ")
vetorResultado[0] = valoresLinha[0]
vetorLinha = list(map(float,valoresLinha[1:]))
vetorResultado[1] = DTW(vetorLinha,self.vetordeteste).peso() #Retorna o peso do caminho
matrizDistancia.append(vetorResultado)
return matrizDistancia
def get_dtw_alignement(intvl1, intvl2):
feats1, feats2 = map(features_getter.get_features, [intvl1, intvl2])
distance_array = cosine_distance(feats1, feats2)
try:
cost, _, paths = DTW(feats1,
feats2,
return_alignment=True,
dist_array=distance_array)
except:
print('\n\n'.join(
str(x) for x in [
feats1.shape, feats2.shape, distance_array.shape, intvl1,
intvl2
]))
path1, path2 = paths[1:]
assert len(path1) == len(path2)
return path1, path2
def classifySeq(self, input_seq):
# print('input seq:', input_seq)
self.class_labels = np.copy(self.c_labels)
self.distances = np.zeros((len(self.X)), dtype=float)
dtw = DTW()
dtw.setY(input_seq)
for i, seq in enumerate(self.X):
# print('\tcomparing with', i)
dtw.setX(seq)
self.distances[i] = dtw.perform_dtw()
bubble_sort(self.K, self.distances, self.class_labels)
self.bins = np.bincount(self.class_labels[:self.K])
self.assigned_classlabel = self.bins.argmax()
# print('assigned class label: ', self.assigned_classlabel)
return self.assigned_classlabel
def main():
context = zmq.Context()
socket: zmq.Socket = context.socket(zmq.PUB)
done: zmq.Socket = context.socket(zmq.SUB)
done.subscribe('')
done.setsockopt(zmq.RCVTIMEO, 2000)
socket.connect(DATA_ENDPOINT)
done.connect(COMMAND_ENDPOINT)
predictors = []
models, false_model = get_models(sys.argv[1])
for model in models:
x = [recording for recording, _ in model]
sample_len = int(
np.ceil(np.average([recording_len for _, recording_len in model])))
predictor = DTW(dtw_cost, sample_len)
predictor.compile(dtw_k)
predictor.train(x, [[0.0, 1.0] for _ in range(len(model))])
predictors.append(predictor)
def cb(found):
socket.send_string('true' if found else 'false')
print('found!', found)
audio_handler = AudioHandler(cb, predictors)
audio_handler.start()
signal_received = False
def signal_cb(_, __):
nonlocal signal_received
signal_received = True
signal.signal(signal.SIGINT, signal_cb)
print('started')
socket.send_string('started')
while not signal_received:
try:
string = done.recv_string()
if string == 'done':
signal_received = True
break
except zmq.ZMQError as err:
print('Error while receiving: ', err)
time.sleep(1)
audio_handler.stop()
socket.close()
done.close()
def one_sym_fn(A, B):
return DTW.dist_damn_adv(A, B, [(-1, -2), (-1, -1), (-2, -1)], [[2, 1], [2], [2, 1]])
same_spkrs = 0
diff_spkrs = 0
with open("SysE.zs") as rf:
cword = ''
fs = []
for line in rf:
l = line.rstrip('\n')
if l == '':
continue
if "Class" in line:
cword = l.split()[1]
fs = []
else:
fname, start, end = l.split()
start = int(float(start) * FBANKS_RATE)
end = int(float(end) * FBANKS_RATE)
tmp = do_fbank(fname)[start:end+1]
for (fname2, tmp2) in fs:
dtw = DTW(tmp, tmp2, return_alignment=1)
spkr1 = fname[:3]
spkr2 = fname2[:3]
if spkr1 == spkr2:
same_spkrs += 1
else:
diff_spkrs += 1
pairs.append((cword, spkr1, spkr2, tmp, tmp2, dtw[0], dtw[-1][1], dtw[-1][2]))
fs.append((fname, tmp))
joblib.dump(pairs, "from_aren.joblib",
compress=3, cache_size=512)
print "ratio same spkrs / all:", float(same_spkrs) / (same_spkrs + diff_spkrs)
def normal_asym_fn(A, B):
return DTW.dist_damn_adv(A, B, [(0, -1), (-1, -1), (-1, 0)], [[1], [1], [1]])
"train": File.open_with_label("Beef_TRAIN"),
"test": File.open_with_label("Beef_TEST"),
"fn": two_sym_fn,
},
{
"title": "Beef.P=2.asym",
"train": File.open_with_label("Beef_TRAIN"),
"test": File.open_with_label("Beef_TEST"),
"fn": two_asym_fn,
},
]
for i, job in enumerate(jobs):
print("job:", i, "of", len(jobs))
print("job:", job["title"])
result = {"title": job["title"]}
start_time = time.process_time()
accuracy, correctness = DTW.predict_list(job["train"], job["test"], job["fn"])
end_time = time.process_time()
result["accuracy"] = accuracy
result["correctness"] = correctness
result["total"] = len(job["test"])
result["time_elapsed"] = end_time - start_time
File.write_json(job["title"] + ".json", result)
def do_dtw(x1, x2):
dtw = DTW(x1, x2, return_alignment=1)
return dtw[0], dtw[-1][1], dtw[-1][2]
for i in xrange(len(source_list)):
target = STF()
target.loadfile(target_list[i])
mfcc = MFCC(target.SPEC.shape[1] * 2, target.frequency, dimension = DIMENSION)
target_mfcc = numpy.array([mfcc.mfcc(target.SPEC[frame]) for frame in xrange(target.SPEC.shape[0])])
target_data = numpy.hstack([target_mfcc, mfcc.delta(target_mfcc)])
source = STF()
source.loadfile(source_list[i])
mfcc = MFCC(source.SPEC.shape[1] * 2, source.frequency)
source_mfcc = numpy.array([mfcc.mfcc(source.SPEC[frame]) for frame in xrange(source.SPEC.shape[0])])
dtw = DTW(source_mfcc, target_mfcc, window = abs(source.SPEC.shape[0] - target.SPEC.shape[0]) * 2)
warp_mfcc = dtw.align(source_mfcc)
warp_data = numpy.hstack([warp_mfcc, mfcc.delta(warp_mfcc)])
data = numpy.hstack([warp_data, target_data])
if learn_data is None:
learn_data = data
else:
learn_data = numpy.vstack([learn_data, data])
square_mean = (square_mean * (learn_data.shape[0] - target_mfcc.shape[0]) + (target_mfcc ** 2).sum(axis = 0)) / learn_data.shape[0]
mean = (mean * (learn_data.shape[0] - target_mfcc.shape[0]) + target_mfcc.sum(axis = 0)) / learn_data.shape[0]
gmm = sklearn.mixture.GMM(n_components = 2, covariance_type = 'full')
gmm.fit(learn_data)
}
# using Euclidean
print('using Euclidean...')
correctness = 0
for i, each in enumerate(testing):
print(i, "of", len(testing))
label = each[0]
data = each[1:]
min_dist = float('Inf')
min_pos = -1
for j, candidate in enumerate(training_with_label):
dist = DTW.euclid_dist(data, candidate['data'])
if dist < min_dist:
min_dist = dist
min_pos = j
predicted = training_with_label[min_pos]['label']
if predicted == label:
correctness += 1
print('min_dist:', min_dist)
print('min_pos:', min_pos)
print('predicted label:', predicted)
print('actual label:', label)
def do_dtw_pair(p1, p2):
dtw = DTW(p1[2], p2[2], return_alignment=1)
# word, talkerX, talkerY, x, y, cost_dtw, dtw_x_to_y_mapping, dtw_y_to_x_mapping
return p1[0], p1[1], p2[1], p1[2], p2[2], dtw[0], dtw[-1][1], dtw[-1][2]
abc = [a, b, c]
zeros = abc.count(0)
if zeros == 3:
continue
if zeros == 2:
if b == 0:
continue
print('c:', c)
start_time = time.process_time()
accuracy, correctness = DTW.predict_list(job['train'], job['test'], [a, b, c], [ [-1, 0], [0, -1], [-1, -1] ])
print('[', a, b, c, ']', 'accuracy:', accuracy, 'correctness:', correctness)
end_time = time.process_time()
time_elasped = end_time - start_time
print('time_elapsed:', time_elasped)
result = {
'a': a,
'b': b,
'c': c,
'accuracy': accuracy,
'correctness': correctness,
'total': len(job['test']),
'time_elapsed': time_elasped
def two_asym_fn(A, B):
return DTW.dist_damn_adv(A, B, [(-2, -3), (-1, -1), (-3, -2)], [[2 / 3, 2 / 3, 2 / 3], [1], [1, 1, 1]])
def one_asym_fn(A, B):
return DTW.dist_damn_adv(A, B, [(-1, -2), (-1, -1), (-2, -1)], [[1 / 2, 1 / 2], [1], [1, 1]])
def main():
X = np.array(
[[
-1.870935e-01, -1.941152e+01, -5.882407e+00, -6.821157e+00,
-1.960176e+01, -9.122212e+00, -5.398831e+00, 1.076720e+01,
6.568266e+00, -4.519125e+00, 3.895056e+00, 5.133921e+00,
8.817558e-01, -3.344995e-01, -5.670179e-01, 1.977534e-01,
-2.825252e+00, 5.670564e-02, -4.054463e-01, -1.570377e+00,
2.622198e-01, -1.543853e+00, -2.045755e-01, -3.527825e-01,
-9.646030e-01, 2.183463e-02, -1.032675e-01, -2.037469e-01,
6.428432e-02, 3.759445e-01, 1.421797e-01, 6.430891e-01,
4.654990e-01, 3.725418e-01, 1.287469e-01, -1.014837e-01,
-2.791396e-01, -1.542485e-01, 2.186694e-03
],
[
-1.019952e+00, -1.993099e+01, -4.057678e+00, -1.526845e+01,
-2.177533e+01, -1.245389e+01, -9.609365e+00, 1.133379e+01,
1.232076e+00, -6.754516e+00, 2.665905e+00, 2.272804e+00,
9.301512e-01, -4.683660e-01, -9.745451e-01, 2.767194e-01,
-2.455450e+00, 5.393829e-01, 9.000364e-01, -8.480091e-01,
1.262016e+00, -1.286797e+00, -2.833939e-01, -7.601787e-01,
-1.068123e+00, 3.039335e-02, -1.688788e-01, -1.540181e-01,
1.790668e-01, 9.829438e-01, 1.804260e-01, 8.561022e-01,
8.271446e-01, 4.591559e-01, 3.011147e-01, -1.383719e-01,
-4.674556e-01, -2.046140e-01, -3.690498e-04
],
[
-1.443142e+00, -2.198701e+01, -5.805995e+00, -1.672378e+01,
-1.823138e+01, -9.483617e+00, -1.114543e+01, 1.179495e+01,
1.517057e+00, -4.424240e+00, 2.745863e+00, 1.741368e+00,
9.667338e-01, -7.838881e-01, -1.382024e+00, 4.797013e-01,
-1.130435e+00, 5.262859e-01, 2.157257e+00, 3.958666e-01,
1.625053e+00, -1.028640e+00, -6.725525e-01, -1.544723e+00,
-1.684104e+00, 2.848946e-02, -1.810844e-01, 7.951448e-02,
2.598753e-01, 1.174168e+00, 6.566963e-02, 4.949601e-01,
8.592530e-01, 3.002176e-01, 3.606367e-01, -2.307723e-01,
-4.826202e-01, -2.030101e-01, -4.236597e-03
],
[
-1.900802e+00, -2.299645e+01, -4.536950e+00, -1.414699e+01,
-1.758996e+01, -4.441329e+00, -6.765486e+00, 1.656342e+01,
2.659910e+00, -5.983455e+00, 6.687205e-01, 1.489522e+00,
9.912346e-01, -9.541904e-01, -9.296216e-01, 9.521492e-01,
1.242009e+00, 7.240354e-01, 2.593670e+00, 1.582148e+00,
1.876561e+00, -2.959121e-01, -6.624181e-01, -2.094018e+00,
-1.627936e+00, 1.666238e-02, -6.888758e-02, 3.390231e-01,
3.715679e-02, 6.889561e-01, -7.894406e-02, -1.904353e-01,
3.151930e-01, -2.784935e-01, 3.556462e-01, -9.185118e-02,
-8.689702e-02, 1.021361e-01, -7.807227e-03
],
[
-3.666164e+00, -2.478892e+01, -3.244268e+00, -1.303400e+01,
-1.906302e+01, -2.341908e+00, -4.841555e+00, 1.627757e+01,
7.112659e-01, -8.267406e+00, -2.829990e+00, -2.894825e+00,
9.936614e-01, -9.970276e-01, -1.918873e-01, 1.159399e+00,
1.196834e+00, 2.927031e-01, 1.222421e+00, 1.510809e+00,
1.455974e+00, -2.361124e-01, -1.168944e+00, -2.098960e+00,
-1.699771e+00, 7.517332e-03, 1.816315e-01, 3.902620e-01,
-3.455281e-01, -2.452208e-01, -1.617719e-01, -7.106320e-01,
-5.168278e-01, -8.029260e-01, 1.800197e-01, 7.171424e-02,
3.953680e-01, 5.738304e-01, -7.401578e-03
],
[
-4.679300e+00, -2.317812e+01, -5.777994e-01, -1.090323e+01,
-1.773932e+01, -3.056363e+00, -4.850668e+00, 1.847520e+01,
1.554637e-01, -8.145046e+00, -5.016462e+00, -3.548824e+00,
1.000000e+00, -7.062657e-01, 1.255049e-01, 1.227021e-01,
-1.742893e-01, 2.614480e-01, 4.152198e-01, 1.704610e-01,
-4.592960e-02, 9.517462e-02, -4.944558e-01, -9.176090e-01,
-5.496670e-01, 1.843350e-03, 4.214767e-01, 2.275055e-01,
-5.783656e-01, -8.267415e-01, -2.690008e-01, -6.347027e-01,
-9.392050e-01, -1.088750e+00, -6.555431e-02, 2.584541e-01,
8.261142e-01, 8.615100e-01, -5.349388e-03
],
[
-5.039043e+00, -2.285567e+01, -1.988579e+00, -1.236157e+01,
-1.669317e+01, -4.063885e+00, -4.548844e+00, 1.811908e+01,
1.588668e+00, -9.188303e+00, -4.906452e+00, -4.238289e+00,
9.999387e-01, 3.074194e-04, 4.170531e-02, -8.331995e-01,
-1.648457e+00, -5.129647e-02, -3.066546e-01, -1.482461e+00,
-1.428380e+00, -3.240738e-01, -3.979310e-01, -1.560897e-01,
6.459251e-01, -1.109232e-03, 5.914558e-01, 1.447010e-01,
-3.698424e-01, -2.614930e-01, -2.379599e-01, -3.071502e-01,
-8.654104e-01, -9.751623e-01, -2.588414e-01, 3.338255e-01,
9.100384e-01, 8.658913e-01, -4.769736e-03
],
[
-4.745595e+00, -2.333554e+01, -4.551358e+00, -1.535441e+01,
-1.746766e+01, -1.504144e+00, -6.059576e+00, 1.541317e+01,
2.697018e+00, -7.995371e+00, -2.880983e+00, -5.871685e-01,
9.973097e-01, 6.545821e-01, 9.114037e-02, -9.433860e-01,
-1.469063e+00, -4.489590e-01, 1.847529e-01, -1.617276e+00,
-2.124964e+00, -5.797045e-01, 2.443566e-01, 1.065121e+00,
1.506727e+00, -5.771380e-03, 5.974856e-01, 3.222021e-01,
1.184963e-01, 7.826490e-01, -7.407951e-02, -4.043330e-01,
-3.859515e-01, -5.287699e-01, -3.860550e-01, -1.046112e-01,
4.986993e-01, 3.908327e-01, -6.098467e-03
],
[
-3.631593e+00, -2.450166e+01, -5.423533e+00, -1.905079e+01,
-1.945535e+01, -4.651355e+00, -1.164923e+01, 1.066649e+01,
-2.179882e+00, -1.033172e+01, -4.678272e+00, -1.146063e+00,
9.894610e-01, 1.279829e+00, 5.487746e-01, -1.567750e-01,
5.368466e-01, -5.418938e-01, -1.980120e-01, -1.922416e+00,
-2.380239e+00, -1.192883e+00, 1.307755e-01, 1.459865e+00,
1.601482e+00, -1.252421e-02, 3.835591e-01, 4.835010e-01,
3.622566e-01, 1.399534e+00, 1.932966e-01, -4.817923e-01,
1.580537e-01, -4.482182e-02, -1.587683e-01, -4.263003e-01,
7.625756e-02, -4.941015e-02, -6.395079e-03
],
[
-2.110161e+00, -2.189944e+01, -3.577273e+00, -1.490368e+01,
-1.860307e+01, -1.838648e+00, -9.386616e+00, 1.157644e+01,
-8.588143e-01, -6.351494e+00, 1.952272e-01, 2.438842e+00,
9.763833e-01, 1.641409e+00, 1.482953e+00, 3.769910e-01,
2.646360e+00, 1.363502e-01, -1.660754e+00, -1.539307e+00,
-2.213823e+00, -1.400694e+00, -1.281913e+00, 7.679644e-01,
9.266631e-01, -2.294159e-02, 6.459616e-02, 3.304029e-01,
2.195285e-01, 9.616777e-01, 3.012913e-01, -6.065879e-01,
2.680732e-01, 2.129893e-01, 4.940115e-03, -6.909447e-01,
-4.549990e-01, -3.945748e-01, -3.847875e-03
],
[
4.239744e-02, -2.082980e+01, -3.259567e+00, -9.902978e+00,
-1.883495e+01, -4.886611e+00, -1.249734e+01, 8.136093e+00,
-2.597832e+00, -9.356349e+00, 8.548814e-01, 2.256215e+00,
9.477813e-01, 1.424646e+00, 1.763293e+00, 3.178869e-01,
3.291507e+00, 6.225432e-01, -1.792925e+00, -7.311083e-01,
-1.608112e+00, -7.074323e-01, -1.766296e+00, 3.738186e-01,
6.888947e-01, -2.449871e-02, -1.485645e-01, 8.271299e-02,
-3.127924e-02, 2.166191e-01, 2.373298e-01, -4.095209e-01,
2.766629e-01, 2.636068e-01, 1.393205e-01, -5.765115e-01,
-5.688236e-01, -4.159141e-01, -9.079840e-04
],
[
1.624431e+00, -1.775662e+01, -3.748471e+00, -6.696738e+00,
-1.709612e+01, -9.690301e+00, -1.333205e+01, 5.609426e+00,
-4.097603e+00, -1.489259e+01, -1.807804e+00, 2.345169e+00,
9.034401e-01, 9.051237e-01, 1.135900e+00, -8.310343e-02,
1.962012e+00, 4.752693e-01, -2.050676e+00, -8.725550e-01,
-1.446113e+00, -7.977447e-01, -2.261846e+00, -6.668284e-01,
-9.852156e-03, -1.902324e-02, -1.992105e-01, -1.321529e-01,
-1.321248e-01, -2.698160e-01, 5.306010e-02, -1.037717e-01,
1.192216e-01, 1.697413e-01, 1.115534e-01, -2.455495e-01,
-3.910235e-01, -2.571701e-01, 1.331323e-03
]])
Y = np.array(
[[
-1.870935e-01, -1.941152e+01, -5.882407e+00, -6.821157e+00,
-1.960176e+01, -9.122212e+00, -5.398831e+00, 1.076720e+01,
6.568266e+00, -4.519125e+00, 3.895056e+00, 5.133921e+00,
8.817558e-01, -3.344995e-01, -5.670179e-01, 1.977534e-01,
-2.825252e+00, 5.670564e-02, -4.054463e-01, -1.570377e+00,
2.622198e-01, -1.543853e+00, -2.045755e-01, -3.527825e-01,
-9.646030e-01, 2.183463e-02, -1.032675e-01, -2.037469e-01,
6.428432e-02, 3.759445e-01, 1.421797e-01, 6.430891e-01,
4.654990e-01, 3.725418e-01, 1.287469e-01, -1.014837e-01,
-2.791396e-01, -1.542485e-01, 2.186694e-03
],
[
-1.019952e+00, -1.993099e+01, -4.057678e+00, -1.526845e+01,
-2.177533e+01, -1.245389e+01, -9.609365e+00, 1.133379e+01,
1.232076e+00, -6.754516e+00, 2.665905e+00, 2.272804e+00,
9.301512e-01, -4.683660e-01, -9.745451e-01, 2.767194e-01,
-2.455450e+00, 5.393829e-01, 9.000364e-01, -8.480091e-01,
1.262016e+00, -1.286797e+00, -2.833939e-01, -7.601787e-01,
-1.068123e+00, 3.039335e-02, -1.688788e-01, -1.540181e-01,
1.790668e-01, 9.829438e-01, 1.804260e-01, 8.561022e-01,
8.271446e-01, 4.591559e-01, 3.011147e-01, -1.383719e-01,
-4.674556e-01, -2.046140e-01, -3.690498e-04
],
[
-1.443142e+00, -2.198701e+01, -5.805995e+00, -1.672378e+01,
-1.823138e+01, -9.483617e+00, -1.114543e+01, 1.179495e+01,
1.517057e+00, -4.424240e+00, 2.745863e+00, 1.741368e+00,
9.667338e-01, -7.838881e-01, -1.382024e+00, 4.797013e-01,
-1.130435e+00, 5.262859e-01, 2.157257e+00, 3.958666e-01,
1.625053e+00, -1.028640e+00, -6.725525e-01, -1.544723e+00,
-1.684104e+00, 2.848946e-02, -1.810844e-01, 7.951448e-02,
2.598753e-01, 1.174168e+00, 6.566963e-02, 4.949601e-01,
8.592530e-01, 3.002176e-01, 3.606367e-01, -2.307723e-01,
-4.826202e-01, -2.030101e-01, -4.236597e-03
],
[
-1.900802e+00, -2.299645e+01, -4.536950e+00, -1.414699e+01,
-1.758996e+01, -4.441329e+00, -6.765486e+00, 1.656342e+01,
2.659910e+00, -5.983455e+00, 6.687205e-01, 1.489522e+00,
9.912346e-01, -9.541904e-01, -9.296216e-01, 9.521492e-01,
1.242009e+00, 7.240354e-01, 2.593670e+00, 1.582148e+00,
1.876561e+00, -2.959121e-01, -6.624181e-01, -2.094018e+00,
-1.627936e+00, 1.666238e-02, -6.888758e-02, 3.390231e-01,
3.715679e-02, 6.889561e-01, -7.894406e-02, -1.904353e-01,
3.151930e-01, -2.784935e-01, 3.556462e-01, -9.185118e-02,
-8.689702e-02, 1.021361e-01, -7.807227e-03
],
[
-3.666164e+00, -2.478892e+01, -3.244268e+00, -1.303400e+01,
-1.906302e+01, -2.341908e+00, -4.841555e+00, 1.627757e+01,
7.112659e-01, -8.267406e+00, -2.829990e+00, -2.894825e+00,
9.936614e-01, -9.970276e-01, -1.918873e-01, 1.159399e+00,
1.196834e+00, 2.927031e-01, 1.222421e+00, 1.510809e+00,
1.455974e+00, -2.361124e-01, -1.168944e+00, -2.098960e+00,
-1.699771e+00, 7.517332e-03, 1.816315e-01, 3.902620e-01,
-3.455281e-01, -2.452208e-01, -1.617719e-01, -7.106320e-01,
-5.168278e-01, -8.029260e-01, 1.800197e-01, 7.171424e-02,
3.953680e-01, 5.738304e-01, -7.401578e-03
],
[
-4.679300e+00, -2.317812e+01, -5.777994e-01, -1.090323e+01,
-1.773932e+01, -3.056363e+00, -4.850668e+00, 1.847520e+01,
1.554637e-01, -8.145046e+00, -5.016462e+00, -3.548824e+00,
1.000000e+00, -7.062657e-01, 1.255049e-01, 1.227021e-01,
-1.742893e-01, 2.614480e-01, 4.152198e-01, 1.704610e-01,
-4.592960e-02, 9.517462e-02, -4.944558e-01, -9.176090e-01,
-5.496670e-01, 1.843350e-03, 4.214767e-01, 2.275055e-01,
-5.783656e-01, -8.267415e-01, -2.690008e-01, -6.347027e-01,
-9.392050e-01, -1.088750e+00, -6.555431e-02, 2.584541e-01,
8.261142e-01, 8.615100e-01, -5.349388e-03
],
[
-5.039043e+00, -2.285567e+01, -1.988579e+00, -1.236157e+01,
-1.669317e+01, -4.063885e+00, -4.548844e+00, 1.811908e+01,
1.588668e+00, -9.188303e+00, -4.906452e+00, -4.238289e+00,
9.999387e-01, 3.074194e-04, 4.170531e-02, -8.331995e-01,
-1.648457e+00, -5.129647e-02, -3.066546e-01, -1.482461e+00,
-1.428380e+00, -3.240738e-01, -3.979310e-01, -1.560897e-01,
6.459251e-01, -1.109232e-03, 5.914558e-01, 1.447010e-01,
-3.698424e-01, -2.614930e-01, -2.379599e-01, -3.071502e-01,
-8.654104e-01, -9.751623e-01, -2.588414e-01, 3.338255e-01,
9.100384e-01, 8.658913e-01, -4.769736e-03
],
[
-4.745595e+00, -2.333554e+01, -4.551358e+00, -1.535441e+01,
-1.746766e+01, -1.504144e+00, -6.059576e+00, 1.541317e+01,
2.697018e+00, -7.995371e+00, -2.880983e+00, -5.871685e-01,
9.973097e-01, 6.545821e-01, 9.114037e-02, -9.433860e-01,
-1.469063e+00, -4.489590e-01, 1.847529e-01, -1.617276e+00,
-2.124964e+00, -5.797045e-01, 2.443566e-01, 1.065121e+00,
1.506727e+00, -5.771380e-03, 5.974856e-01, 3.222021e-01,
1.184963e-01, 7.826490e-01, -7.407951e-02, -4.043330e-01,
-3.859515e-01, -5.287699e-01, -3.860550e-01, -1.046112e-01,
4.986993e-01, 3.908327e-01, -6.098467e-03
],
[
-3.631593e+00, -2.450166e+01, -5.423533e+00, -1.905079e+01,
-1.945535e+01, -4.651355e+00, -1.164923e+01, 1.066649e+01,
-2.179882e+00, -1.033172e+01, -4.678272e+00, -1.146063e+00,
9.894610e-01, 1.279829e+00, 5.487746e-01, -1.567750e-01,
5.368466e-01, -5.418938e-01, -1.980120e-01, -1.922416e+00,
-2.380239e+00, -1.192883e+00, 1.307755e-01, 1.459865e+00,
1.601482e+00, -1.252421e-02, 3.835591e-01, 4.835010e-01,
3.622566e-01, 1.399534e+00, 1.932966e-01, -4.817923e-01,
1.580537e-01, -4.482182e-02, -1.587683e-01, -4.263003e-01,
7.625756e-02, -4.941015e-02, -6.395079e-03
],
[
-2.110161e+00, -2.189944e+01, -3.577273e+00, -1.490368e+01,
-1.860307e+01, -1.838648e+00, -9.386616e+00, 1.157644e+01,
-8.588143e-01, -6.351494e+00, 1.952272e-01, 2.438842e+00,
9.763833e-01, 1.641409e+00, 1.482953e+00, 3.769910e-01,
2.646360e+00, 1.363502e-01, -1.660754e+00, -1.539307e+00,
-2.213823e+00, -1.400694e+00, -1.281913e+00, 7.679644e-01,
9.266631e-01, -2.294159e-02, 6.459616e-02, 3.304029e-01,
2.195285e-01, 9.616777e-01, 3.012913e-01, -6.065879e-01,
2.680732e-01, 2.129893e-01, 4.940115e-03, -6.909447e-01,
-4.549990e-01, -3.945748e-01, -3.847875e-03
],
[
4.239744e-02, -2.082980e+01, -3.259567e+00, -9.902978e+00,
-1.883495e+01, -4.886611e+00, -1.249734e+01, 8.136093e+00,
-2.597832e+00, -9.356349e+00, 8.548814e-01, 2.256215e+00,
9.477813e-01, 1.424646e+00, 1.763293e+00, 3.178869e-01,
3.291507e+00, 6.225432e-01, -1.792925e+00, -7.311083e-01,
-1.608112e+00, -7.074323e-01, -1.766296e+00, 3.738186e-01,
6.888947e-01, -2.449871e-02, -1.485645e-01, 8.271299e-02,
-3.127924e-02, 2.166191e-01, 2.373298e-01, -4.095209e-01,
2.766629e-01, 2.636068e-01, 1.393205e-01, -5.765115e-01,
-5.688236e-01, -4.159141e-01, -9.079840e-04
],
[
1.624431e+00, -1.775662e+01, -3.748471e+00, -6.696738e+00,
-1.709612e+01, -9.690301e+00, -1.333205e+01, 5.609426e+00,
-4.097603e+00, -1.489259e+01, -1.807804e+00, 2.345169e+00,
9.034401e-01, 9.051237e-01, 1.135900e+00, -8.310343e-02,
1.962012e+00, 4.752693e-01, -2.050676e+00, -8.725550e-01,
-1.446113e+00, -7.977447e-01, -2.261846e+00, -6.668284e-01,
-9.852156e-03, -1.902324e-02, -1.992105e-01, -1.321529e-01,
-1.321248e-01, -2.698160e-01, 5.306010e-02, -1.037717e-01,
1.192216e-01, 1.697413e-01, 1.115534e-01, -2.455495e-01,
-3.910235e-01, -2.571701e-01, 1.331323e-03
]])
my_warper = DTW()
my_warper.setX(X)
my_warper.setY(Y)
my_warper.perform_dtw()
print(my_warper.get_dtw_dist())
def two_sym_fn(A, B):
return DTW.dist_damn_adv(A, B, [(-2, -3), (-1, -1), (-3, -2)], [[2, 2, 1], [2], [2, 2, 1]])
print("End of creation of the training dictionnary")
keywords = open(folder_unprocessed + "/task/keywords.txt").read().splitlines()
validation_dict = data.create_set('data/test')
transcriptions_dict = data.get_transcriptions(folder_unprocessed + '/ground-truth/transcription.txt')
print("Number of element in the validation dict: %d" % len(validation_dict))
output_lines = []
for keyword in keywords:
data = keyword.split(',')
output_line = [data[0]]
score_dict = {}
image = training_dict[data[1]]
dtw_o = DTW(image)
i = 0
for key, image2 in validation_dict.items():
score = dtw_o.calculate_cost(image2)
score_dict[key] = score
i += 1
if i % 100 == 0:
print(i)
sorted_score = dict(sorted(score_dict.items(), key = operator.itemgetter(1)))
for name, score in sorted_score.items():
output_line.append(name)
output_line.append(score)
output_lines.append(output_line)
# for key, image in validation_dict.items():
# dtw_o = DTW(image)
def bad_fn(A, B):
return DTW.dist_damn_adv(A, B, [(-1, -1)], [[1]])
def do_dtw(word, x, y):
dtw = DTW(x, y, return_alignment=1)
# word, x, y, cost_dtw, dtw_x_to_y_mapping, dtw_y_to_x_mapping
return word, x, y, dtw[0], dtw[-1][1], dtw[-1][2]
