The deadline for this position has now passed. Please do not apply.


PhD Research Opportunity TU Dublin School of Media
Automatic Updating of Crowd-sourced Maps Using Deep Learning

Application Deadline: June 30, 2019
PhD Start Date: September 2019

Objectives of project: 
Pattern matching in raster digital imagery using machine learning has received a lot of recent attention but automatic updating of crowd-sourced vector maps using these techniques is still an open problem. Today, VGI maps like OSM are well populated with built environment data – how to keep them up-to-date in a timely manner is the main research question, with plenty of scope to develop new techniques for automating this process.

To address this problem, we propose a novel change/damage detection system using satellite imagery and deep learning techniques (multi-layered neural networks) in order to automate the updating of freely available online crowd-sourced maps like OpenStreetMap (OSM). We ensure the quality of information added by our automated system is up to standard by using inter-rater reliability statistics. Stipend and academic fee waiver.

The successful applicant will receive a stipend of €16,000 per annum through TU Dublin’s College of Arts & Tourism Postgraduate Research Scholarship Scheme for the project’s duration (up to a maximum of four years).

The cost of EU academic fees will be covered for the duration of the project (subject to successful annual evaluations). Non-EU applicants are eligible to take up the project; however, they will be required to fund the fee differential.

The aforementioned stipend and EU academic fee waiver constitute part of a scholarship, and not part of a contract of employment. The successful candidate will be a student, rather than an employee of TU Dublin.

Candidates will be either:
• a Masters graduate in computer science, spatial information science, or related geospatial discipline awarded at 2.1 or higher; and/or,
• an outstanding (minimum upper 2:1) Bachelors in computer science, spatial information science, or related geospatial discipline with evidence of high quality research potential.

• Background, knowledge or experience in spatial databases, OSM Editing/Overpass API.
• Background, knowledge or experience in image processing, GIS application development.
• Background, knowledge or experience in machine learning.

How to apply: 
Completed applications and requests for further information submitted directly to Dr. James D. Carswell by email at

A complete application includes:

i) A full curriculum vitae – including qualifications, experience, list of publications

ii) A personal statement outlining motivations for pursuing PhD research (maximum 500 words), and,

iii) A sample of the candidate’s written work (these will not be returned).

Incomplete applications will not be considered

///Advert posted 17th June, 2019.