Spaces:
Runtime error
Runtime error
Upload 3 files
Browse files- app.py +141 -0
- hashies.txt +1 -0
- requirements.txt +11 -0
app.py
ADDED
@@ -0,0 +1,141 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from PIL import Image
|
3 |
+
import numpy as np
|
4 |
+
import nltk
|
5 |
+
nltk.download('stopwords')
|
6 |
+
nltk.download('punkt')
|
7 |
+
import pandas as pd
|
8 |
+
import random
|
9 |
+
import easyocr
|
10 |
+
import re
|
11 |
+
from nltk.corpus import stopwords
|
12 |
+
from nltk.tokenize import word_tokenize
|
13 |
+
from sklearn.feature_extraction.text import TfidfVectorizer
|
14 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
15 |
+
from transformers import AutoTokenizer, VisionEncoderDecoderModel, ViTFeatureExtractor
|
16 |
+
@st.cache(allow_output_mutation=True)
|
17 |
+
|
18 |
+
# Directory path to the saved model on Google Drive
|
19 |
+
model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
|
20 |
+
|
21 |
+
# Load the feature extractor and tokenizer
|
22 |
+
feature_extractor = ViTFeatureExtractor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
|
23 |
+
tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
|
24 |
+
|
25 |
+
|
26 |
+
def generate_captions(image):
|
27 |
+
generated_caption = tokenizer.decode(model.generate(feature_extractor(image, return_tensors="pt").pixel_values.to("cpu"))[0])
|
28 |
+
sentence = generated_caption
|
29 |
+
text_to_remove = "<|endoftext|>"
|
30 |
+
generated_caption = sentence.replace(text_to_remove, "")
|
31 |
+
return generated_caption
|
32 |
+
|
33 |
+
# use easyocr to extract text from the image
|
34 |
+
def image_text(image):
|
35 |
+
reader = easyocr.Reader(['en'])
|
36 |
+
text = reader.readtext(np.array(image))
|
37 |
+
detected_text = " ".join([item[1] for item in text])
|
38 |
+
|
39 |
+
# Extract individual words, convert to lowercase, and add "#" symbol
|
40 |
+
detected_text= ['#' + entry[1].strip().lower().replace(" ", "_") for entry in text]
|
41 |
+
return detected_text
|
42 |
+
|
43 |
+
# Load NLTK stopwords for filtering
|
44 |
+
stop_words = set(stopwords.words('english'))
|
45 |
+
|
46 |
+
# Add hashtags to keywords, which have been generated from image captioing
|
47 |
+
def add_hashtags(keywords):
|
48 |
+
hashtags = []
|
49 |
+
|
50 |
+
for keyword in keywords:
|
51 |
+
# Generate hashtag from the keyword (you can modify this part as per your requirements)
|
52 |
+
hashtag = '#' + keyword.lower()
|
53 |
+
|
54 |
+
hashtags.append(hashtag)
|
55 |
+
|
56 |
+
return hashtags
|
57 |
+
|
58 |
+
def trending_hashtags():
|
59 |
+
# Read trending hashtags from a file separated by commas
|
60 |
+
with open("hashies.txt", "r") as file:
|
61 |
+
hashtags_string = file.read()
|
62 |
+
|
63 |
+
# Split the hashtags by commas and remove any leading/trailing spaces
|
64 |
+
trending_hashtags = [hashtag.strip() for hashtag in hashtags_string.split(',')]
|
65 |
+
|
66 |
+
# Create a DataFrame from the hashtags
|
67 |
+
df = pd.DataFrame(trending_hashtags, columns=["Hashtags"])
|
68 |
+
|
69 |
+
# Function to extract keywords from a given text
|
70 |
+
def extract_keywords(text):
|
71 |
+
tokens = word_tokenize(text)
|
72 |
+
keywords = [token.lower() for token in tokens if token.lower() not in stop_words]
|
73 |
+
return keywords
|
74 |
+
|
75 |
+
# Extract keywords from caption and trending hashtags
|
76 |
+
caption_keywords = extract_keywords(caption)
|
77 |
+
hashtag_keywords = [extract_keywords(hashtag) for hashtag in df["Hashtags"]]
|
78 |
+
|
79 |
+
# Function to calculate cosine similarity between two strings
|
80 |
+
def calculate_similarity(text1, text2):
|
81 |
+
tfidf_vectorizer = TfidfVectorizer()
|
82 |
+
tfidf_matrix = tfidf_vectorizer.fit_transform([text1, text2])
|
83 |
+
similarity_matrix = cosine_similarity(tfidf_matrix[0], tfidf_matrix[1])
|
84 |
+
return similarity_matrix[0][0]
|
85 |
+
|
86 |
+
# Calculate similarity between caption and each trending hashtag
|
87 |
+
similarities = [calculate_similarity(' '.join(caption_keywords), ' '.join(keywords)) for keywords in hashtag_keywords]
|
88 |
+
|
89 |
+
# Sort trending hashtags based on similarity in descending order
|
90 |
+
sorted_hashtags = [hashtag for _, hashtag in sorted(zip(similarities, df["Hashtags"]), reverse=True)]
|
91 |
+
|
92 |
+
# Select top k relevant hashtags (e.g., top 5) without duplicates
|
93 |
+
selected_hashtags = list(set(sorted_hashtags[:5]))
|
94 |
+
|
95 |
+
selected_hashtag = [word.strip("'") for word in selected_hashtags]
|
96 |
+
|
97 |
+
return selected_hashtag
|
98 |
+
|
99 |
+
# create the Streamlit app
|
100 |
+
def app():
|
101 |
+
st.title('Iamge from your Side, Trending Hashtags from our Site')
|
102 |
+
|
103 |
+
st.write('Upload an image to see what we have in store, Alwyas For You!')
|
104 |
+
|
105 |
+
# create file uploader
|
106 |
+
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])
|
107 |
+
|
108 |
+
# check if file has been uploaded
|
109 |
+
if uploaded_file is not None:
|
110 |
+
# load the image
|
111 |
+
image = Image.open(uploaded_file)
|
112 |
+
|
113 |
+
# Image Captions
|
114 |
+
string = generate_captions(uploaded_file)
|
115 |
+
tokens = word_tokenize(string)
|
116 |
+
keywords = [token.lower() for token in tokens if token.lower() not in stop_words]
|
117 |
+
hashtags = add_hashtags(keywords)
|
118 |
+
|
119 |
+
# Text Captions from image
|
120 |
+
in_image_text = image_text(uploaded_file)
|
121 |
+
|
122 |
+
#Final Hashtags Generation
|
123 |
+
web_hashtags = trending_hashtags()
|
124 |
+
|
125 |
+
combined_hashtags = hashtags + in_image_text + web_hashtags
|
126 |
+
|
127 |
+
# Shuffle the list randomly
|
128 |
+
random.shuffle(combined_hashtags)
|
129 |
+
|
130 |
+
combined_hashtags = list(set(item for item in combined_hashtags[:15] if not re.search(r'\d$', item)))
|
131 |
+
|
132 |
+
|
133 |
+
# display the image
|
134 |
+
st.image(image, width=500, height=400)
|
135 |
+
st.write("Here it is THE CAPTIONS Just! for your Photo",string)
|
136 |
+
st.write("ohh! finally we have decided these are the best ",combined_hashtags)
|
137 |
+
|
138 |
+
# run the app
|
139 |
+
if __name__ == '__main__':
|
140 |
+
app()
|
141 |
+
|
hashies.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
'#instagood','#sand','#surfboard ','#love', '#photography', '#instagram', '#photooftheday', '#india', '#picoftheday', '#nature', '#instadaily', '#likeforlikes', '#follow', '#fashion', '#travel', '#followforfollowback', '#kerala', '#beautiful', '#style', '#art', '#travelphotography', '#followme', '#model', '#naturephotography', '#photo', '#smile', '#instalike', '#happy', '#bhfyp', '#like', '#life', '#stayhome', '#mumbai', '#cute', '#photoshoot', '#photographer', '#friends', '#me', '#instamood', '#food', '#likeforfollow', '#travelgram', '#lifestyle', '#keralagram', '#insta', '#trending', '#beauty', '#fun', '#like4like', '#igers', '#tiktok', '#like4likes', '#sunset', '#incredibleindia', '#throwback', '#girl', '#staysafe', '#keralatourism', '#likeforlike', '#bestoftheday', '#fashionblogger', '#kochi', '#quarantine', '#sky', '#portrait', '#wanderlust', '#keralagodsowncountry', '#tbt', '#likes', '#instafashion', '#instapic', '#followers', '#mobilephotography', '#music', '#fitness', '#mallu', '#delhi', '#swag', '#photographers_of_india', '#amazing', '#motivation', '#follow4follow', '#explore', '#godsowncountry', '#summer', '#followforfollow', '#modeling', '#loveyourself', '#indianphotography', '#indian', '#mountains', '#landscape', '#look', '#likesforlike', '#goodvibes', '#streetphotography', '#lockdown', '#travelblogger', '#selfie', '#malayalam', '#comment','#nyc', '#dogsofinstagram', '#california', '#cell', '#phone', '#newyork', '#losangeles', '#miami', '#family', '#florida', '#beach', '#foodie', '#adventure', '#explorepage', '#hiking', '#dog', '#foodporn', '#2020', '#makeup', '#socialdistancing', '#artist', '#newyorkcity', '#realestate', '#design', '#puppy', '#covid19', '#ootd', '#blacklivesmatter', '#roadtrip', '#interiordesign', '#outdoors', '#repost', '#blessed', '#colorado', '#vacation', '#texas', '#chicago', '#coronavirus', '#puppiesofinstagram', '#inspiration', '#quarantinelife', '#home', '#puppylove', '#flowers', '#dogstagram', '#nofilter', '#selfcare', '#selflove', '#sandiego', '#getoutside', '#fitnessmotivation', '#entrepreneur', '#smallbusiness', '#dogs', '#france', '#paris', '#sun', '#confinement', '#weekend', '#instamoment', '#sea', '#holidays', '#architecture', '#southoffrance', '#picture', '#vacances', '#marseille', '#igersfrance', '#bretagne', '#mood', '#lyon', '#frenchgirl', '#soleil', '#blackandwhite', '#cotedazur', '#provence', '#amour', '#frenchriviera', '#sunnyday', '#french', '#paysage', '#bordeaux', '#view', '#sport', '#europe', '#plage', '#naturelovers', '#instafood', '#instatravel', '#shooting', '#photographie', '#summervibes', '#nice', '#parisienne', '#trip', '#landscapephotography', '#outfit', '#blue', '#pictureoftheday', '#parisfrance', '#maldives', '#maldivesislands', '#ocean', '#paradise', '#holiday', '#indianocean', '#island','#islandlife', '#traveltheworld', '#maldivesresorts', '#beautifuldestinations', '#beachlife', '#traveling', '#travelling', '#honeymoon', '#maldiveslovers', '#luxurylifestyle', '#luxury', '#maldive', '#traveler', '#malediven', '#traveller', '#beautifulmaldives', '#luxurytravel', '#memories', '#maldivas', '#diving', '#scubadiving', '#tropical', '#relax', '#beachvibes', '#maldivesisland', '#underwaterphotography', '#sunnysideoflife', '#underwater', '#traveladdict', '#couplegoals', '#maldivesmania', '#resort', '#travelblog', '#maldivesinsider', '#clouds', '#sand'
|
requirements.txt
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
streamlit==0.88.0
|
2 |
+
Pillow==8.4.0
|
3 |
+
numpy==1.21.4
|
4 |
+
nltk==3.6.5
|
5 |
+
pandas==1.3.4
|
6 |
+
easyocr==1.4
|
7 |
+
scikit-learn==0.24.2
|
8 |
+
torch==1.9.0
|
9 |
+
transformers==4.11.3
|
10 |
+
altair<5
|
11 |
+
click<8.0
|