Authors: Mohammed Shaiqur Rahman, Ibne Farabi Shihab, Lynna Chu, Anuj Sharma
Published on: April 18, 2024
Impact Score: 8.0
Arxiv code: Arxiv:2404.12258
Summary
- What is new: Introduction of DeepLocalization, a system combining Graph-Based Change-Point Detection with a Video-LLM for driver behavior monitoring.
- Why this is important: The issue of distracted driving and its contribution to road accidents.
- What the research proposes: A novel framework leveraging both change-point detection for action localization and video large language models for activity classification.
- Results: Achieved 57.5% accuracy in event classification and 51% in event detection on the SynDD2 dataset.
Technical Details
Technological frameworks used: DeepLocalization, including Graph-Based Change-Point Detection and Video Large Language Model (Video-LLM)
Models used: Customized Video-LLM with prompt engineering
Data used: SynDD2 dataset
Potential Impact
Automotive safety systems, driver monitoring system manufacturers, and auto insurance companies.
Want to implement this idea in a business?
We have generated a startup concept here: SafeDriveAI.
Leave a Reply