industry 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 Risk State Prediction at Construction Sites VLM-based risk prediction system for detecting unsafe worker behavior and equipment hazards — funded by Doosan Enerbility Zero-Shot Captioning for Driver Status Reporting Zero-shot image captioning for in-vehicle driver monitoring without task-specific labeled data — funded by Hyundai NGV
