Conference paper / 2026
Multi-Frame Grid Perspective for Traffic Video Captioning and Context-Aware VQA
Proceedings of AI for Accelerated Research Symposium · EPiC Series in Technology, 3, 65–76
Research summary
This study examines how arrangements of video frames can help vision-language models interpret traffic scenes. It compares visual treatments for object location, attention, and movement, connecting visual representation with captioning and question answering.