Update app.py
Browse files
app.py
CHANGED
@@ -27,44 +27,320 @@ from transformers import CLIPProcessor, CLIPModel
|
|
27 |
########################################
|
28 |
|
29 |
landmark_titles = [
|
30 |
-
"Eiffel Tower",
|
31 |
-
"
|
32 |
-
"
|
33 |
-
"
|
34 |
-
"
|
35 |
-
|
36 |
-
"
|
37 |
-
"
|
38 |
-
"
|
39 |
-
"
|
40 |
-
"Stonehenge",
|
41 |
-
|
42 |
-
"
|
43 |
-
"
|
44 |
-
"
|
45 |
-
"
|
46 |
-
"
|
47 |
-
|
48 |
-
"
|
49 |
-
"
|
50 |
-
"
|
51 |
-
"
|
52 |
-
"
|
53 |
-
|
54 |
-
"
|
55 |
-
"
|
56 |
-
"
|
57 |
-
"
|
58 |
-
"
|
59 |
-
|
60 |
-
"
|
61 |
-
"
|
62 |
-
"
|
63 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
64 |
]
|
65 |
|
|
|
66 |
def clean_text(text):
|
67 |
-
# Очистка от некоторых спецсимволов, ссылок, диакритики
|
68 |
text = re.sub(r'МФА:?\s?\[.*?\]', '', text)
|
69 |
text = re.sub(r'\[.*?\]', '', text)
|
70 |
|
|
|
27 |
########################################
|
28 |
|
29 |
landmark_titles = [
|
30 |
+
"Eiffel Tower",
|
31 |
+
"Louvre Museum",
|
32 |
+
"Cathédrale Notre-Dame de Paris",
|
33 |
+
"Arc de Triomphe",
|
34 |
+
"Palace of Versailles",
|
35 |
+
|
36 |
+
"Big Ben",
|
37 |
+
"London Eye",
|
38 |
+
"Tower of London",
|
39 |
+
"Buckingham Palace",
|
40 |
+
"Stonehenge",
|
41 |
+
|
42 |
+
"Colosseum",
|
43 |
+
"Pantheon",
|
44 |
+
"Trevi Fountain",
|
45 |
+
"Leaning Tower of Pisa",
|
46 |
+
"Doge's Palace",
|
47 |
+
|
48 |
+
"Sagrada Família",
|
49 |
+
"Alhambra",
|
50 |
+
"Park Güell",
|
51 |
+
"Cathedral of Santiago de Compostela",
|
52 |
+
"La Rambla",
|
53 |
+
|
54 |
+
"Neuschwanstein Castle",
|
55 |
+
"Brandenburg Gate",
|
56 |
+
"Reichstag Building",
|
57 |
+
"Cologne Cathedral",
|
58 |
+
"Hofbräuhaus München",
|
59 |
+
|
60 |
+
"Acropolis of Athens",
|
61 |
+
"Parthenon",
|
62 |
+
"Temple of Olympian Zeus",
|
63 |
+
"Delphi Ruins",
|
64 |
+
"Meteora Monasteries",
|
65 |
+
|
66 |
+
"Charles Bridge",
|
67 |
+
"Prague Castle",
|
68 |
+
"Old Town Square in Prague",
|
69 |
+
"St. Vitus Cathedral",
|
70 |
+
"Cesky Krumlov Castle",
|
71 |
+
|
72 |
+
"Schönbrunn Palace",
|
73 |
+
"Hofburg Palace",
|
74 |
+
"St. Stephen's Cathedral in Vienna",
|
75 |
+
"Belvedere Palace",
|
76 |
+
"Mozarthaus Vienna",
|
77 |
+
|
78 |
+
"Rijksmuseum",
|
79 |
+
"Van Gogh Museum",
|
80 |
+
"Anne Frank House",
|
81 |
+
"Kinderdijk Windmills",
|
82 |
+
"Keukenhof Gardens",
|
83 |
+
|
84 |
+
"Bruges' Historic Centre",
|
85 |
+
"Grand Place in Brussels",
|
86 |
+
"Atomium",
|
87 |
+
"Ghent Belfry",
|
88 |
+
"Basilica of the Holy Blood",
|
89 |
+
|
90 |
+
"Great Wall of China",
|
91 |
+
"Forbidden City",
|
92 |
+
"Terracotta Army",
|
93 |
+
"Potala Palace",
|
94 |
+
"Temple of Heaven",
|
95 |
+
|
96 |
+
"Taj Mahal",
|
97 |
+
"Red Fort",
|
98 |
+
"Gateway of India",
|
99 |
+
"Hawa Mahal",
|
100 |
+
"Golden Temple",
|
101 |
+
|
102 |
+
"Tokyo Tower",
|
103 |
+
"Fushimi Inari-taisha",
|
104 |
+
"Kiyomizu-dera",
|
105 |
+
"Himeji Castle",
|
106 |
+
"Itsukushima Shrine",
|
107 |
+
|
108 |
+
"Petronas Twin Towers",
|
109 |
+
"Batu Caves",
|
110 |
+
"Langkawi Sky Bridge",
|
111 |
+
"Kek Lok Si Temple",
|
112 |
+
"Mount Kinabalu",
|
113 |
+
|
114 |
+
"Gardens by the Bay",
|
115 |
+
"Marina Bay Sands",
|
116 |
+
"Merlion Park",
|
117 |
+
"Buddha Tooth Relic Temple",
|
118 |
+
"Chinatown Singapore",
|
119 |
+
|
120 |
+
"Angkor Wat",
|
121 |
+
"Ta Prohm",
|
122 |
+
"Bayon Temple",
|
123 |
+
"Preah Khan",
|
124 |
+
"Banteay Srei",
|
125 |
+
|
126 |
+
"Borobudur Temple",
|
127 |
+
"Prambanan Temple",
|
128 |
+
"Tanah Lot",
|
129 |
+
"Uluwatu Temple",
|
130 |
+
"Mount Bromo",
|
131 |
+
|
132 |
+
"Ha Long Bay",
|
133 |
+
"Imperial City of Huế",
|
134 |
+
"My Son Sanctuary",
|
135 |
+
"Hoi An Ancient Town",
|
136 |
+
"Phong Nha-Ke Bang National Park",
|
137 |
+
|
138 |
+
"Gyeongbokgung Palace",
|
139 |
+
"Bukchon Hanok Village",
|
140 |
+
"N Seoul Tower",
|
141 |
+
"Jeju Island",
|
142 |
+
"Changdeokgung Palace",
|
143 |
+
|
144 |
+
"Shwedagon Pagoda",
|
145 |
+
"Bagan Temples",
|
146 |
+
"Inle Lake",
|
147 |
+
"Kyaiktiyo Pagoda",
|
148 |
+
"Mandalay Palace",
|
149 |
+
|
150 |
+
"Pyramids of Giza",
|
151 |
+
"Great Sphinx of Giza",
|
152 |
+
"Karnak Temple",
|
153 |
+
"Valley of the Kings",
|
154 |
+
"Abu Simbel Temples",
|
155 |
+
|
156 |
+
"Victoria Falls",
|
157 |
+
"Hwange National Park",
|
158 |
+
"Matobo Hills",
|
159 |
+
"Great Zimbabwe Ruins",
|
160 |
+
"Lake Kariba",
|
161 |
+
|
162 |
+
"Serengeti National Park",
|
163 |
+
"Mount Kilimanjaro",
|
164 |
+
"Ngorongoro Crater",
|
165 |
+
"Zanzibar Stone Town",
|
166 |
+
"Lake Manyara",
|
167 |
+
|
168 |
+
"Table Mountain",
|
169 |
+
"Robben Island",
|
170 |
+
"Kruger National Park",
|
171 |
+
"Cape of Good Hope",
|
172 |
+
"Blyde River Canyon",
|
173 |
+
|
174 |
+
"Djmaa el Fna",
|
175 |
+
"Koutoubia Mosque",
|
176 |
+
"Hassan II Mosque",
|
177 |
+
"Chefchaouen",
|
178 |
+
"Aït Benhaddou",
|
179 |
+
|
180 |
+
"Gorée Island",
|
181 |
+
"Lake Retba (Lac Rose)",
|
182 |
+
"Saloum Delta",
|
183 |
+
"Saint-Louis Island",
|
184 |
+
"Niokolo-Koba National Park",
|
185 |
+
|
186 |
+
"Sossusvlei",
|
187 |
+
"Etosha National Park",
|
188 |
+
"Fish River Canyon",
|
189 |
+
"Skeleton Coast",
|
190 |
+
"Twyfelfontein",
|
191 |
+
|
192 |
+
"Lalibela Churches",
|
193 |
+
"Simien Mountains",
|
194 |
+
"Blue Nile Falls",
|
195 |
+
"Aksum Obelisks",
|
196 |
+
"Harar Jugol",
|
197 |
+
|
198 |
+
"Carthage Archaeological Site",
|
199 |
+
"El Jem Amphitheatre",
|
200 |
+
"Medina of Tunis",
|
201 |
+
"Sidi Bou Said",
|
202 |
+
"Dougga",
|
203 |
+
|
204 |
+
"Mount Kenya",
|
205 |
+
"Maasai Mara National Reserve",
|
206 |
+
"Lake Nakuru",
|
207 |
+
"Amboseli National Park",
|
208 |
+
"Tsavo National Park",
|
209 |
+
|
210 |
+
"Statue of Liberty",
|
211 |
+
"Grand Canyon",
|
212 |
+
"Yellowstone National Park",
|
213 |
+
"Mount Rushmore",
|
214 |
+
"Golden Gate Bridge",
|
215 |
+
|
216 |
+
"Niagara Falls",
|
217 |
+
"CN Tower",
|
218 |
+
"Banff National Park",
|
219 |
+
"Notre-Dame Basilica in Montreal",
|
220 |
+
"Stanley Park",
|
221 |
+
|
222 |
+
"Chichén Itzá",
|
223 |
+
"Teotihuacan",
|
224 |
+
"Tulum Ruins",
|
225 |
+
"Copper Canyon",
|
226 |
+
"Palenque",
|
227 |
+
|
228 |
+
"Havana Old Town",
|
229 |
+
"Varadero Beach",
|
230 |
+
"Trinidad Historic Center",
|
231 |
+
"Viñales Valley",
|
232 |
+
"Castillo de San Pedro de la Roca",
|
233 |
+
|
234 |
+
"Altos de Chavón",
|
235 |
+
"Santo Domingo Colonial Zone",
|
236 |
+
"Punta Cana Beaches",
|
237 |
+
"Los Haitises National Park",
|
238 |
+
"Bahía de las Águilas",
|
239 |
+
|
240 |
+
"Blue Mountains",
|
241 |
+
"Dunn's River Falls",
|
242 |
+
"Bob Marley Museum",
|
243 |
+
"Negril Seven Mile Beach",
|
244 |
+
"Port Royal",
|
245 |
+
|
246 |
+
"Belize Barrier Reef",
|
247 |
+
"Caracol",
|
248 |
+
"Lamanai",
|
249 |
+
"Xunantunich",
|
250 |
+
"Caye Caulker",
|
251 |
+
|
252 |
+
"Tikal",
|
253 |
+
"Antigua Guatemala",
|
254 |
+
"Lake Atitlán",
|
255 |
+
"Semuc Champey",
|
256 |
+
"Pacaya Volcano",
|
257 |
+
|
258 |
+
"Panama Canal",
|
259 |
+
"Casco Viejo",
|
260 |
+
"San Blas Islands",
|
261 |
+
"Bocas del Toro",
|
262 |
+
"Coiba National Park",
|
263 |
+
|
264 |
+
"Arenal Volcano",
|
265 |
+
"Manuel Antonio National Park",
|
266 |
+
"Monteverde Cloud Forest",
|
267 |
+
"Tortuguero National Park",
|
268 |
+
"Corcovado National Park",
|
269 |
+
|
270 |
+
"Machu Picchu",
|
271 |
+
"Nazca Lines",
|
272 |
+
"Lake Titicaca",
|
273 |
+
"Colca Canyon",
|
274 |
+
"Cusco Historic Center",
|
275 |
+
|
276 |
+
"Red Square",
|
277 |
+
"Saint Basil's Cathedral",
|
278 |
+
"Kremlin (Moscow)",
|
279 |
+
"Hermitage Museum",
|
280 |
+
"Peterhof",
|
281 |
+
"Catherine Palace",
|
282 |
+
"Lake Baikal",
|
283 |
+
"Valley of Geysers (Kamchatka)",
|
284 |
+
"Kazan Kremlin",
|
285 |
+
"Kizhi Pogost",
|
286 |
+
|
287 |
+
"Christ the Redeemer",
|
288 |
+
"Sugarloaf Mountain",
|
289 |
+
"Iguazu Falls",
|
290 |
+
"Amazon Rainforest",
|
291 |
+
"Copacabana Beach",
|
292 |
+
|
293 |
+
"Easter Island (Rapa Nui)",
|
294 |
+
"Torres del Paine",
|
295 |
+
"Valle de la Luna",
|
296 |
+
"Atacama Desert",
|
297 |
+
"San Cristobal Hill",
|
298 |
+
|
299 |
+
"Perito Moreno Glacier",
|
300 |
+
"Mount Fitz Roy",
|
301 |
+
"Buenos Aires Obelisk",
|
302 |
+
"La Boca Neighborhood",
|
303 |
+
"Talampaya National Park",
|
304 |
+
|
305 |
+
"Salt Cathedral of Zipaquirá",
|
306 |
+
"Cartagena Old Town",
|
307 |
+
"Tayrona National Park",
|
308 |
+
"Cocora Valley",
|
309 |
+
"Monserrate",
|
310 |
+
|
311 |
+
"Galápagos Islands",
|
312 |
+
"Cotopaxi Volcano",
|
313 |
+
"Quilotoa Lake",
|
314 |
+
"Middle of the World City",
|
315 |
+
"Historic Center of Quito",
|
316 |
+
|
317 |
+
"Los Roques Archipelago",
|
318 |
+
"Angel Falls",
|
319 |
+
"Margarita Island",
|
320 |
+
"Canaima National Park",
|
321 |
+
"Roraima",
|
322 |
+
|
323 |
+
"Salar de Uyuni",
|
324 |
+
"Laguna Colorada",
|
325 |
+
"Tiwanaku",
|
326 |
+
"La Paz Cable Car",
|
327 |
+
"Death Road (Yungas Road)",
|
328 |
+
|
329 |
+
"Casapueblo",
|
330 |
+
"Punta del Este",
|
331 |
+
"Colonia del Sacramento",
|
332 |
+
"Plaza Independencia in Montevideo",
|
333 |
+
"Teatro Solís",
|
334 |
+
|
335 |
+
"Itaipu Dam",
|
336 |
+
"Asunción Historic Centre",
|
337 |
+
"Jesuit Missions of La Santísima Trinidad",
|
338 |
+
"Ybycuí National Park",
|
339 |
+
"Cerro Tobatí"
|
340 |
]
|
341 |
|
342 |
+
|
343 |
def clean_text(text):
|
|
|
344 |
text = re.sub(r'МФА:?\s?\[.*?\]', '', text)
|
345 |
text = re.sub(r'\[.*?\]', '', text)
|
346 |
|