Image-to-Hazard: GPT-based Logic Reasoning for Hazard Identification in Construction Site using CCTV Data

This paper is accepted by the 41st International Symposium on Automation and Robotics in Construction (ISARC), Lille, France, June 3 - 7 , 2024. Link to access paper.

Proposed approach

Demo

Sample results

Pose estimation

Convert image coordinate to global coordinate

Hazard identification with GPT-based models

Acknowledgement

This work is inspired by Image2Paragraph, GPT API, OpenMMLab.