Back to search

Georgia Institute of Technology

Curve Sign Inventorying Method Using Smartphones and Deep Learning Technologies

Abstract

dc:description.abstract

The objective of the proposed research is to develop and assess a system using smartphones and deep learning technologies to automatically establish an intelligent and sustainable curve sign inventory from videos. The Manual on the Uniform Traffic Control Devices (MUTCD) is the nationwide regulator that defines the standards used for transportation asset installation and maintenance. The proposed system is one of the components of a larger methodology whose purpose is to accomplish a frequent and cost-effective MUTCD curve sign compliance checking and other curve safety checking in order to reduce the number of deadly crashes on curves. To automatically build an effective sign inventory from videos, four modules are needed: sign detection, classification, tracking and localization. For this purpose, a pipeline has been developed in the past by former students of the Transportation laboratory of Georgia Tech. However, this pipeline is not accurate enough and its different modules have never been critically tested and assessed. Therefore, the objective of this study is to improve the different modules and particularly the detection module, which is the most important module of the pipeline, and to critically assess these improved modules to determine the pipeline ability to build an effective sign inventory. The proposed system has been tested and assessed in real conditions on a mountain road with many curves and curve signs; it has shown that the detection module is able to detect every single curve sign with a very low number of detected non-curve signs (false positive), resulting in a precision of 0.97 and a recall of 1. The other modules also showed very promising results. Overall, this study demonstrates that the proposed system is suitable for building an accurate curve sign inventory that can be used by transportation agencies to get a precise idea of the condition of the curve sign networks on a particular road.

Degree

thesis:*
Level thesis:degree_level
Masters
Department dc:contributor.department
Electrical and Computer Engineering
Grantor dc:publisher
Georgia Institute of Technology
Year dc:date.issued
2022

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Francois-Marchal, Marius Michel
Advisors dc:contributor.advisor
  • Yezzi, Anthony
  • Tsai, Yi-Chang James
Committee member dc:contributor.committeemember
  • AlRegib, Ghassan

Subjects

dc:subject × 8

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
http://hdl.handle.net/1853/70163

Chain of custody

source
Harvested from
Georgia Tech
Base URL
repository.gatech.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Francois-Marchal, Marius Michel. Curve Sign Inventorying Method Using Smartphones and Deep Learning Technologies. Masters thesis, Georgia Institute of Technology, 2022. http://hdl.handle.net/1853/70163