def demo():
if request.method == 'POST':
path = request.form['filename']
file = path
fileObj = feature_extraction.extractFeatures(file)
var_sum, means_sum = fileObj.means()
strObj = "{},{}".format(means_sum, var_sum)
with open('testData', 'w') as fl:
fl.write(strObj)
val = subprocess.check_output(["octave", "classifyMusic.m"])
val = float(val[657:])
if(val>=0.5):
ans = "Genre: Metal"
else:
ans = "Genre: Classical"
return render_template('demo.html', ans=ans)
return render_template('demo.html',ans="")
def extractFeaturesImages(imgs):
# Create a list to append feature vectors to
imagesFeatures = []
# Iterate through the list of images
for img in range(0, len(imgs)):
# Read in each one by one
image = imgs[img]
hog_feat = configParams['use_hog_feat']
spatial_feat = configParams['use_spatial_feat']
hist_feat = configParams['use_hist_feat']
imgFeatures = extractFeatures(image, verbose=False, hog_feat=hog_feat, spatial_feat=spatial_feat, hist_feat=hist_feat)
imagesFeatures.append(imgFeatures)
# Return list of feature vectors
return imagesFeatures
def search_windows(img, windows, classifier, X_scaler):
hog_feat = configParams['use_hog_feat']
spatial_feat = configParams['use_spatial_feat']
hist_feat = configParams['use_hist_feat']
# Create an empty list to receive positive detection windows
on_windows = []
count_window = 0
count_car_window = 0
# Iterate over all windows in the list
for window in windows:
# Extract the test window from original image
window_img = cv2.resize(img[window[0][1]:window[1][1],
window[0][0]:window[1][0]],
(64, 64))
# Extract features for that window
img_features = extractFeatures(window_img, verbose=False,
hog_feat=hog_feat, spatial_feat=spatial_feat, hist_feat=hist_feat)
# Scale extracted features to be fed to classifier
test_features = X_scaler.transform(np.array(img_features).reshape(1, -1))
# Predict using your classifier
prediction = classifier.predict(test_features)
# If positive (prediction == 1) then save the window
count_window += 1
if prediction == 1:
count_car_window += 1
on_windows.append(window)
# Return windows for positive detections
return on_windows
import feature_extraction
'''This file is for the generation of the dataSet for classification'''
file = 'classical/classical.00057.wav'
fileObj = feature_extraction.extractFeatures(file)
var_sum, means_sum = fileObj.means()
print("{},{}".format(means_sum, var_sum))
def SSG_LUGIA(sequence_fasta_file_path=None,
genome_sequence=None,
model_name=None,
model_parameters=None):
"""
SSG-LUGIA Pipeline
Keyword Arguments:
sequence_fasta_file_path {string} -- path to the genome sequence fasta file (default: {None})
genome_sequence {string} -- the actual genome sequence (default: {None})
model_name {string} -- name of one of the standard models or path to model json file (default: {None})
model_parameters {dictionary} -- SSG-LUGIA model configuration (default: {None})
Returns:
list of tuples : list of the tuples containing genomic islands start and end coordinates
"""
if (sequence_fasta_file_path !=
None): # if a genome sequence fasta file has been provided
genome = loadGenome(sequence_fasta_file_path)
elif (genome_sequence !=
None): # otherwise the genome sequence has been provided
genome = genome_sequence
else: # no genome sequence has been provided
raise ValueError(
'Please input either genome sequence of sequence fasta file path')
if (model_name != None):
# if a standard model name has been provided
if (model_name in ['SSG-LUGIA-F', 'SSG-LUGIA-R', 'SSG-LUGIA-P']):
model = loadModel(model_name)
# if path to the json file has been provided
else:
model = loadModelJson(model_name)
elif (model_parameters !=
None): # a custom model parameter configuration has been provided
model = model_parameters
else: # interactively input the model
model = inputModel()
X = extractFeatures(genome, model) # extracting features
print()
print('Detecting Anomalies')
(yp, ys) = detectAnomalies(X, model) # Detecting anomalies
print('Post-Processing')
ys_mf = medianFiltering(ys, model) # applying median filter
yp_mf = binaizeFilteredDistance(ys_mf) # binarizing the values
label_pred = inferLabel(yp_mf, len(genome), model) # infer prediction
label = trimRegions(label_pred, model) # eliminate small islands
genomic_islands = getGenomicIslands(label) # extract genomic islands
print('Total Number of Genomic Islands = {}'.format(len(genomic_islands)))
print('Genomic Islands :')
for genomic_island in genomic_islands:
print('{}\t{}'.format(genomic_island[0], genomic_island[1]))
return genomic_islands # return the islands