def task5b(self, id, t):
given_image_index = self.data_ids.index(float(id))
res = self.lsh_index.query(self.data[given_image_index])
print('With repetition Overall considered images= ',len(res))
res = set(res)
print('Total unique considered images= ',len(res))
'''
for i in res:
print("index =",i)
print("data id =",self.data_ids[i])
print("value =",np.linalg.norm(self.data[i]-self.data[given_image_index]))
'''
''' To Rank them in order '''
distances = sorted_list(t, 'distance', True)
for i in res:
dist = np.linalg.norm(self.data[i]-self.data[given_image_index])
distances.add({'id': self.data_ids[i], 'distance': dist})
print('Top 5 similar images and similarity score using LSH')
print()
pic_info = []
print('Image id - Distance')
for i in range(0, t):
o = distances.extract()
print(str(int(o['id']))+' - '+str(o['distance']))
pic_info.append({'id': str(int(o['id'])), 'info': str(int(o['id']))})
PA.display_images(pic_info, 'Task 5b - '+str(id)+' - '+str(t))
'''To delete'''
def task_2b(self, data, c):
c_a = Clustering_Algorithms()
clusters = c_a.normalised_cut(data, c)
for i in clusters:
pic_info = []
print(len(i))
for j in i:
pic_info.append({'id': data.img_ids[j], 'info': ''})
PA.display_images(pic_info, 'Task 2b - ' + str(c))
def task_2a(self, data, c):
c_a = Clustering_Algorithms()
labels = c_a.spectral_clustering(data, c)
clusters = [[] for _ in range(c)]
for i in range(len(labels)):
clusters[labels[i]].append({'id': data.img_ids[i], 'info': ''})
for i in range(c):
print(len(clusters[i]))
PA.display_images(clusters[i], 'Task 2a - ' + str(c))
def task_3(self, data, k):
pr = PageRanks()
ranks = pr.page_rank(data)
result = ranks.nlargest(k)
pic_info = []
for idx, val in result.iteritems():
pic_info.append({'id': idx, 'info': idx + ' :'+str(val)})
PA.display_images(pic_info, 'Task 3 - '+str(k))
def task6a(self, fileName, k=None):
labelled_data = pd.read_csv(fileName, delim_whitespace=True)
labelled_data = labelled_data.drop(0)
labelled_dict = {}
for index, row in labelled_data.iterrows():
labelled_dict[float(row['image'])] = row['label']
self.process_distances(labelled_dict)
total = 0
for key in self.cluster_dict:
pic_info = [{
'id': str(int(id)),
'info': str(int(id)) + " " + key
} for id in self.cluster_dict[key]]
PA.display_images(pic_info, 'Task 6a - ' + str(fileName))
#print(pic_info)
print("cluster name=", key)
print("images in this cluster= ", self.cluster_dict[key])
print("size of cluster= ", len(self.cluster_dict[key]))
total += len(self.cluster_dict[key])
print("total=", total)
def task6b(self, data, fileName):
labelled_data = pd.read_csv(fileName, delim_whitespace=True)
labelled_data = labelled_data.drop(0)
labelled_dict = {}
for index, row in labelled_data.iterrows():
if row['label'] not in labelled_dict:
labelled_dict[row['label']] = [row['image']]
else:
labelled_dict[row['label']].append(row['image'])
pagelables = []
pageranks = []
for key in labelled_dict:
pr = PageRanks()
ranks = pr.personalized_page_rank(data, labelled_dict[key], False)
pageranks.append(ranks)
pagelables.append(key)
pic_info = []
for i in range(len(data.img_ids)):
max = 0
lables = []
for j in range(len(pagelables)):
if pageranks[j][i] > max:
lables = [pagelables[j]]
max = pageranks[j][i]
elif pageranks[j][i] == max:
lables.append(pagelables[j])
pic_info.append({
'id':
str(data.img_ids[i]),
'info':
str(data.img_ids[i]) + " : " + ",".join(lables)
})
PA.display_images(pic_info, 'Task 6b - ' + str(fileName))