Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import BlipProcessor, BlipForConditionalGeneration
|
| 3 |
+
import torch
|
| 4 |
+
from PIL import Image
|
| 5 |
+
from deep_translator import GoogleTranslator
|
| 6 |
+
from gradio.themes import Base
|
| 7 |
+
|
| 8 |
+
# Load BLIP model and processor
|
| 9 |
+
caption_model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
|
| 10 |
+
caption_processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
|
| 11 |
+
|
| 12 |
+
# Translator
|
| 13 |
+
translator = GoogleTranslator(source='en', target='hi')
|
| 14 |
+
|
| 15 |
+
# Device setup
|
| 16 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 17 |
+
caption_model.to(device)
|
| 18 |
+
|
| 19 |
+
def generate_caption(image):
|
| 20 |
+
try:
|
| 21 |
+
# Preprocess the image
|
| 22 |
+
inputs = caption_processor(images=image, return_tensors="pt")
|
| 23 |
+
pixel_values = inputs.pixel_values.to(device)
|
| 24 |
+
|
| 25 |
+
# Generate caption
|
| 26 |
+
output_ids = caption_model.generate(
|
| 27 |
+
pixel_values,
|
| 28 |
+
max_length=30,
|
| 29 |
+
num_beams=10,
|
| 30 |
+
num_beam_groups=2,
|
| 31 |
+
diversity_penalty=0.5,
|
| 32 |
+
repetition_penalty=2.0,
|
| 33 |
+
temperature=0.6,
|
| 34 |
+
top_k=50,
|
| 35 |
+
top_p=0.95,
|
| 36 |
+
no_repeat_ngram_size=3
|
| 37 |
+
)
|
| 38 |
+
|
| 39 |
+
# Decode English caption
|
| 40 |
+
english_caption = caption_processor.decode(output_ids[0], skip_special_tokens=True)
|
| 41 |
+
|
| 42 |
+
# Translate to Hindi
|
| 43 |
+
hindi_caption = translator.translate(english_caption)
|
| 44 |
+
|
| 45 |
+
return english_caption, hindi_caption
|
| 46 |
+
|
| 47 |
+
except Exception as e:
|
| 48 |
+
return f"Error: {str(e)}", f"Error: {str(e)}"
|
| 49 |
+
|
| 50 |
+
# Custom theme with a blue and white color scheme
|
| 51 |
+
custom_theme = gr.themes.Default(
|
| 52 |
+
primary_hue="blue", # Main color (buttons, highlights)
|
| 53 |
+
secondary_hue="gray", # Secondary elements
|
| 54 |
+
neutral_hue="slate", # Backgrounds, borders
|
| 55 |
+
text_size="lg", # Larger text for readability
|
| 56 |
+
radius_size="md", # Rounded corners
|
| 57 |
+
font=[gr.themes.GoogleFont("Roboto"), "sans-serif"] # Modern font
|
| 58 |
+
)
|
| 59 |
+
|
| 60 |
+
# Gradio interface with improved visuals
|
| 61 |
+
interface = gr.Interface(
|
| 62 |
+
fn=generate_caption,
|
| 63 |
+
inputs=gr.Image(type="pil", label="Upload an Image"),
|
| 64 |
+
outputs=[
|
| 65 |
+
gr.Textbox(label="English Caption", lines=2, placeholder="English caption will appear here..."),
|
| 66 |
+
gr.Textbox(label="Hindi Caption", lines=2, placeholder="हिंदी कैप्शन यहाँ दिखाई देगा...")
|
| 67 |
+
],
|
| 68 |
+
title="Image Caption Generator (English & Hindi)",
|
| 69 |
+
description="Upload an image to generate captions in English and Hindi with a sleek, modern interface.",
|
| 70 |
+
theme=custom_theme,
|
| 71 |
+
css="""
|
| 72 |
+
.gradio-container { max-width: 800px; margin: auto; }
|
| 73 |
+
h1 { text-align: center; color: #1E40AF; }
|
| 74 |
+
.label { font-weight: bold; }
|
| 75 |
+
input, output { border-radius: 8px; }
|
| 76 |
+
"""
|
| 77 |
+
)
|
| 78 |
+
|
| 79 |
+
interface.launch()
|