Fusecap Image Captioning

🚀 FuseCap: Leveraging Large Language Models for Enriched Fused Image Captions

FuseCap is a framework crafted to generate semantically rich image captions, offering a novel approach to image captioning.

🚀 Quick Start

Our BLIP-based model can be run using the following code:

import requests
from PIL import Image
from transformers import BlipProcessor, BlipForConditionalGeneration
import torch

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
processor = BlipProcessor.from_pretrained("noamrot/FuseCap")
model = BlipForConditionalGeneration.from_pretrained("noamrot/FuseCap").to(device)

img_url = 'https://huggingface.co/spaces/noamrot/FuseCap/resolve/main/bike.jpg' 
raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')

text = "a picture of "
inputs = processor(raw_image, text, return_tensors="pt").to(device)

out = model.generate(**inputs, num_beams = 3)
print(processor.decode(out[0], skip_special_tokens=True))

✨ Features

• Semantically Rich Captions: Generate high - quality, semantically rich image captions.
• BLIP - Based Model: Utilize a powerful BLIP - based model for caption generation.

📚 Documentation

Resources

• 💻 Project Page: For more details, visit the official project page.
• 📝 Read the Paper: You can find the paper here.
• 🚀 Demo: Try out our BLIP - based model demo trained using FuseCap.
• 📂 Code Repository: The code for FuseCap can be found in the GitHub repository.
• 🗃️ Datasets: The fused captions datasets can be accessed from here.

Upcoming Updates

The official codebase, datasets and trained models for this project will be released soon.

📄 License

This project is licensed under the MIT license.

📚 BibTeX

@inproceedings{rotstein2024fusecap,
  title={Fusecap: Leveraging large language models for enriched fused image captions},
  author={Rotstein, Noam and Bensa{\"\i}d, David and Brody, Shaked and Ganz, Roy and Kimmel, Ron},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  pages={5689--5700},
  year={2024}
}