File size: 3,574 Bytes
ea32bf7 d3313c5 ea32bf7 922847d ea32bf7 922847d cd4b2c9 922847d ea32bf7 922847d ea32bf7 922847d ea32bf7 922847d ea32bf7 922847d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 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 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 | import gradio as gr
from transformers import BlipProcessor, BlipForConditionalGeneration, MarianMTModel, MarianTokenizer
from PIL import Image
import torch
import random
import datetime
# Load translation model
translator_model_ar = MarianMTModel.from_pretrained("Helsinki-NLP/opus-mt-en-ar")
translator_tokenizer_ar = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-ar")
# Load BLIP model (fine-tuned)
model_path = "Projectt123/MuniVis"
processor_en = BlipProcessor.from_pretrained(model_path)
model_en = BlipForConditionalGeneration.from_pretrained(model_path)
model_en.eval()
# Function to describe image
@torch.no_grad()
def describe_image(image, language):
pil = image if isinstance(image, Image.Image) else Image.fromarray(image)
# وصف بالإنجليزية
inputs = processor_en(pil, return_tensors="pt")
out_ids = model_en.generate(**inputs)
en_desc = processor_en.decode(out_ids[0], skip_special_tokens=True)
# الترجمة إذا اللغة عربية
if str(language).lower().startswith(("ar", "arabic")) or language == "العربية":
inputs_ar = translator_tokenizer_ar(en_desc, return_tensors="pt")
ar_tokens = translator_model_ar.generate(**inputs_ar)
ar_desc = translator_tokenizer_ar.decode(ar_tokens[0], skip_special_tokens=True)
return ar_desc
else:
return en_desc
ASIR_BBOX = {
"lat_min": 17.0, "lat_max": 20.0,
"lon_min": 41.5, "lon_max": 44.0,
}
def random_point_in_bbox(bbox):
lat = random.uniform(bbox["lat_min"], bbox["lat_max"])
lon = random.uniform(bbox["lon_min"], bbox["lon_max"])
return round(lat, 6), round(lon, 6)
def render_html(caption_text, lat, lon, language="English"):
now = datetime.datetime.now()
dt_str = now.strftime("%d %B %Y - %I:%M %p")
is_ar = str(language).lower().startswith(("ar", "arabic")) or language == "العربية"
dt_line = f"🕒 {'التاريخ والوقت' if is_ar else 'Date & Time'}: {dt_str}"
loc_label = "📍 الموقع" if is_ar else "📍 Location"
zoom = 16
osm_iframe = (
f'<iframe width="100%" height="300" frameborder="0" '
f'src="https://www.openstreetmap.org/export/embed.html?layer=mapnik&marker={lat}%2C{lon}&zoom={zoom}"></iframe>'
f'<div style="font-size:12px;color:#aaa;margin-top:4px">{loc_label}: {lat}, {lon} '
f'• <a target="_blank" href="https://www.openstreetmap.org/?mlat={lat}&mlon={lon}#map={zoom}/{lat}/{lon}">Open map</a>'
f'</div>'
)
return f"""
<div style="font-family:system-ui,Segoe UI,Arial;color:#eee;line-height:1.5">
<div style="background:transparent;padding:14px 16px;border-radius:10px;margin-bottom:10px;">
<div style="font-size:16px;white-space:pre-wrap;">{caption_text}</div>
</div>
{osm_iframe}
<div style="margin-top:8px;font-size:14px;">{dt_line}</div>
</div>
"""
def ui_print_like(image, language):
caption = describe_image(image, language)
lat, lon = random_point_in_bbox(ASIR_BBOX)
return render_html(caption, lat, lon, language)
# Gradio UI
demo = gr.Interface(
fn=ui_print_like,
inputs=[
gr.Image(type="pil", label="Upload road image"),
gr.Dropdown(choices=["Arabic", "English", "العربية"], value="English", label="Select Language")
],
outputs=gr.HTML(label="output"),
title="MuniVis – Road Issue Detector (Asir)",
description="Upload an image and get a description with a map (Asir random location) and date/time."
)
demo.launch(debug=True)
|