Retrieval-Augmented Dense Video Captioning & QA
End-to-end pipeline for automated highlight extraction and natural language QA over long-form videos — funded by Piaspace
Funded by Piaspace · 2025
Built an end-to-end pipeline for automated highlight extraction and dense captioning of long-form videos (1hr+), with a retrieval-augmented QA module enabling natural language queries over video content.
Key Contributions
- Designed a scalable pipeline for processing broadcast-length video
- Developed retrieval-augmented QA module for natural language video search
- Deployed on real broadcast data for client demo