CoolFace
Apppublic

binit-bhagat/video-qa

sourceHugging Faceupdated 1y agoView on Hugging Face
0likes
App README

๐ŸŽฅ Multimodal Video Question Answering

An open-source Streamlit app that allows users to ask natural language questions about a video by combining frame analysis with transcript retrieval.

Features

  • โ€”Audio transcription using Whisper
  • โ€”Frame embeddings via CLIP
  • โ€”Text embeddings via Sentence Transformers
  • โ€”Retrieval via FAISS
  • โ€”VQA using SmolVLM

Run

bash
streamlit run streamlit_app.py