MANIKANDAN A
commited on
Commit
·
083d065
1
Parent(s):
7f7a61c
Update app.py
Browse files
app.py
CHANGED
@@ -100,22 +100,58 @@ def login_section():
|
|
100 |
|
101 |
|
102 |
|
103 |
-
def predict(
|
104 |
captions = []
|
105 |
pred_caption = generate_caption('tmp.jpg', caption_model)
|
106 |
|
107 |
cap_col.markdown('#### Predicted Captions:')
|
108 |
-
|
|
|
109 |
|
110 |
for _ in range(4):
|
111 |
pred_caption = generate_caption('tmp.jpg', caption_model, add_noise=True)
|
112 |
if pred_caption not in captions:
|
113 |
-
|
|
|
114 |
|
115 |
cap_col.markdown('<div class="caption-container">', unsafe_allow_html=True)
|
116 |
for c in captions:
|
117 |
cap_col.markdown(f'<div class="cap-line" style="color: black; background-color: light grey; padding: 5px; margin-bottom: 5px; font-family: \'Palatino Linotype\', \'Book Antiqua\', Palatino, serif;">{c}</div>', unsafe_allow_html=True)
|
118 |
cap_col.markdown('</div>', unsafe_allow_html=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
119 |
|
120 |
|
121 |
|
@@ -135,45 +171,9 @@ def main():
|
|
135 |
elif selected_tab == "Login":
|
136 |
login_section()
|
137 |
elif selected_tab == "Generate Caption":
|
138 |
-
|
139 |
-
if hasattr(st.session_state, "username"):
|
140 |
-
st.markdown(f"<p style='{heading_style}'>Generate Caption</p>", unsafe_allow_html=True)
|
141 |
-
st.markdown("Upload an image to generate a caption:")
|
142 |
-
|
143 |
-
with st.sidebar:
|
144 |
-
st.title("Options")
|
145 |
-
selected_languages = st.multiselect("Select languages for translation:", ['en', 'ta', 'hi', 'zh-cn', 'es', 'fr', 'de', 'it', 'ja'])
|
146 |
-
img_url = st.text_input("Enter Image URL:")
|
147 |
-
img_upload = st.file_uploader("Upload Image:", type=['jpg', 'png', 'jpeg'])
|
148 |
-
|
149 |
-
col1, col2 = st.columns(2)
|
150 |
-
if img_url or img_upload:
|
151 |
-
if img_url:
|
152 |
-
img = Image.open(requests.get(img_url, stream=True).raw)
|
153 |
-
else:
|
154 |
-
img = Image.open(img_upload)
|
155 |
-
|
156 |
-
img = img.convert('RGB')
|
157 |
-
col1.image(img, caption="Input Image", use_column_width=True)
|
158 |
-
img.save('tmp.jpg')
|
159 |
-
predict(col2)
|
160 |
-
|
161 |
-
for lang in selected_languages:
|
162 |
-
if lang != "en":
|
163 |
-
translated_caption = translator.translate(generate_caption, src="en", dest=lang)
|
164 |
-
st.markdown(f"<p style='font-size: 24px; font-weight: bold; margin-bottom: 20px;'>{lang.upper()} Translation:</p>", unsafe_allow_html=True)
|
165 |
-
st.write(translated_caption.text)
|
166 |
-
|
167 |
-
# Update the caption in the database
|
168 |
-
|
169 |
-
|
170 |
-
|
171 |
-
|
172 |
else:
|
173 |
st.write("Please login to access this feature.")
|
174 |
|
175 |
-
|
176 |
-
|
177 |
-
|
178 |
if __name__ == "__main__":
|
179 |
main()
|
|
|
100 |
|
101 |
|
102 |
|
103 |
+
def predict():
|
104 |
captions = []
|
105 |
pred_caption = generate_caption('tmp.jpg', caption_model)
|
106 |
|
107 |
cap_col.markdown('#### Predicted Captions:')
|
108 |
+
translated_caption = translate_caption(pred_caption, target_language)
|
109 |
+
captions.append(translated_caption)
|
110 |
|
111 |
for _ in range(4):
|
112 |
pred_caption = generate_caption('tmp.jpg', caption_model, add_noise=True)
|
113 |
if pred_caption not in captions:
|
114 |
+
translated_caption = translate_caption(pred_caption, target_language)
|
115 |
+
captions.append(translated_caption)
|
116 |
|
117 |
cap_col.markdown('<div class="caption-container">', unsafe_allow_html=True)
|
118 |
for c in captions:
|
119 |
cap_col.markdown(f'<div class="cap-line" style="color: black; background-color: light grey; padding: 5px; margin-bottom: 5px; font-family: \'Palatino Linotype\', \'Book Antiqua\', Palatino, serif;">{c}</div>', unsafe_allow_html=True)
|
120 |
cap_col.markdown('</div>', unsafe_allow_html=True)
|
121 |
+
|
122 |
+
st.markdown('<h1 style="text-align:center; font-family:Arial; width:fit-content; font-size:3em; color:black; text-shadow: 2px 2px 4px #000000;">IMAGE CAPTION GENERATOR</h1>', unsafe_allow_html=True)
|
123 |
+
col1, col2 = st.columns(2)
|
124 |
+
|
125 |
+
# Image URL input
|
126 |
+
img_url = st.text_input(label='Enter Image URL')
|
127 |
+
|
128 |
+
# Image upload input
|
129 |
+
img_upload = st.file_uploader(label='Upload Image', type=['jpg', 'png', 'jpeg'])
|
130 |
+
|
131 |
+
# Language selection dropdown
|
132 |
+
target_language = st.selectbox('Select Target Language', ['en', 'ta', 'hi', 'es', 'fr', 'zh-cn'], index=0)
|
133 |
+
|
134 |
+
# Process image and generate captions
|
135 |
+
if img_url:
|
136 |
+
img = Image.open(requests.get(img_url, stream=True).raw)
|
137 |
+
img = img.convert('RGB')
|
138 |
+
col1.image(img, caption="Input Image", use_column_width=True)
|
139 |
+
img.save('tmp.jpg')
|
140 |
+
predict(col2)
|
141 |
+
|
142 |
+
st.markdown('<center style="opacity: 70%">OR</center>', unsafe_allow_html=True)
|
143 |
+
|
144 |
+
elif img_upload:
|
145 |
+
img = img_upload.read()
|
146 |
+
img = Image.open(io.BytesIO(img))
|
147 |
+
img = img.convert('RGB')
|
148 |
+
col1.image(img, caption="Input Image", use_column_width=True)
|
149 |
+
img.save('tmp.jpg')
|
150 |
+
predict(col2)
|
151 |
+
|
152 |
+
# Remove temporary image file
|
153 |
+
if img_url or img_upload:
|
154 |
+
os.remove('tmp.jpg')
|
155 |
|
156 |
|
157 |
|
|
|
171 |
elif selected_tab == "Login":
|
172 |
login_section()
|
173 |
elif selected_tab == "Generate Caption":
|
174 |
+
predict()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
175 |
else:
|
176 |
st.write("Please login to access this feature.")
|
177 |
|
|
|
|
|
|
|
178 |
if __name__ == "__main__":
|
179 |
main()
|