Dec 08, 2023
Summary
Minseok Kim - Rutgers University
Full Description
Title: Interfacing Computer Vision AI/ML in Geospatial Data with Supply Chain Management
Abstract: In this seminar, we will start with explore the acquisition and processing of geo-spatial data, delve into computer vision machine learning tools specific to geo-spatial analysis, and discuss the latest innovations and current trends in AI/ML. The comprehensive study will follow up with presentation of an innovative approach that combines advanced AI techniques with geospatial data analysis for enhanced urban planning and efficient monitoring of global supply chains. Our research introduces a novel stratified segmentation technique, the Stratified Building Analysis Framework, which leverages deep learning and remote sensing for effective urban building analysis. Utilizing a cloud-
augmented multi-scale classification framework (SRMSC), our method simplifies spatial data collection by employing basic building footprints and emulating human reasoning in image interpretation. In parallel, addressing the critical challenge of port congestion impacting global supply chains, we advance an AI-powered method for the remote estimation of cargo container volumes. By integrating a zero-shot segmentation model with Generative Adversarial Networks, our approach enables precise detection and volume estimation of cargo containers through advanced shadow analysis. The synergy of these methods underscores the potential of computer vision AI in transforming urban development and supply chain management, offering substantial advancements for both academic research and industrial applications.
