📷 AI Image Captioning App

A simple Streamlit app that uses the BLIP/BLIP2 models to generate captions for any uploaded image.

🚀 How It Works

1. Upload an image (jpg/png)
2. The AI model will analyze it
3. You’ll get a descriptive caption like:

   “A cat wearing sunglasses while sitting on a couch.”

🧠 Tech Stack

• Python
• Streamlit
• Transformers (BLIP model from Salesforce)
• Hugging Face

🛠️ Run Locally

pip install streamlit transformers torch torchvision Pillow
streamlit run ai_image_captioning_app.py

🧠 Model Comparison: BLIP Base vs Large vs BLIP2

We tested the same image using 3 different models. Here's how their descriptions differ:

🏃 Input Image:

---

⚙️ Using BLIP Base Model

model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")

🔥 Using BLIP Large Model

model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")

🧠 Using BLIP2 + Flan-T5 Model

processor = Blip2Processor.from_pretrained("Salesforce/blip2-flan-t5-xl")
model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-flan-t5-xl")