README.md

metadata

title: Street Descriptions using MuniVis
emoji: 🛣️
colorFrom: indigo
colorTo: blue
sdk: gradio
sdk_version: 5.41.1
app_file: app.py
pinned: false

🛣️ Street Descriptions using MuniVis Model

This Space provides a simple Gradio interface for generating image captions using a fine-tuned BLIP model hosted at saja003/MuniVis. It also includes the option to translate the generated English captions into Arabic using a MarianMT translation model.

---

🔍 Features

• 🧠 Image captioning with a fine-tuned BLIP model.
• 🌐 Language selection: English or Arabic.
• 📸 Upload any street or urban image to generate a descriptive caption.
• 🤖 Translation powered by Helsinki-NLP/opus-mt-en-ar.

---

🚀 How it works

1. Upload an image.
2. Select a language (English or Arabic).
3. Receive a descriptive caption of the scene.
   • If Arabic is selected, the caption is translated automatically.

---

🧠 Models Used

• BLIP (Bootstrapping Language Image Pretraining):
  • Fine-tuned on street and infrastructure images.
  • Model repo: Projectt123/MuniVis
• MarianMT Translation Model:
  • Helsinki-NLP/opus-mt-en-ar for English to Arabic translation.

---

🛠️ Requirements

This Space uses the following libraries:

transformers
torch
gradio
Pillow