A powerful and elegant image captioning system built with the cutting-edge BLIP (Bootstrapping Language-Image Pretraining) and Vision Transformers (ViT) — enabling machines to "see" and describe images just like a human would!
Image credit: Salesforce Research
✅ Generates natural language captions for any image
✅ Leverages Vision Transformers for accurate scene understanding
✅ Built with Hugging Face Transformers, PyTorch, and BLIP
✅ Supports real-time or batch image captioning
✅ Easily extensible for Visual Question Answering or multimodal reasoning
git clone https://github.com/yourusername/blip-image-captioning.git
cd blip-image-captioning
pip install -r requirements.txt
🧠 Model Info
BLIP: A vision-language pretraining framework by Salesforce Research.
Architecture: Uses Vision Transformers (ViT) for image encoding and Transformer decoders for text generation.
Pretrained Models: Downloaded via Hugging Face.