Skip to main content

Banedanmark - Digitisation of cable drawings

From old drawings to vector graphic
Create an innovative digital solution with Banedanmark, Denmark's infrastructure manager

Banedanmark has thousands of cables buried in the ground. The cables are of different sorts, such as live cables and liquid-carrying cables that can be as much as 100 years old.

The information about these cables are drawn - with great variety of quality - into a map and saved in different formats e.g. TIFF, PDF, DWG and GDN and placed in an electronical file system.

Converting these drawings into a Geographical Information System (GIS) is a complicated manual process that is both time consuming and ineffective. Therefore, a new innovative solution needs to be developed.

Banedanmark is launching this crowdsourcing challenge to investigate the possibility of an automated or semi-automated converting process.

If you have any questions, please feel free to reach out to project employee, Kristian Mathiesen on email [email protected] or telephone number +45 2045 2986. 

50.000 kr. / 20.000 kr / 10.000 kr.

The challenges opens.


It is possible to submit solution proposals from 15 November until 15 December. If you or your team have any questions during this period, please do not hesitate to contact Dorte Winther on her email [email protected] or on her phone number +45 20 21 37 36.


All solution proposals are reviewed


The best ideas are selected, and the owners will be invited for the Final Event.

Final Event

The persons with the best solution proposals will participate in the Final Event. It is possible to present via Skype.

BaneDanmark Logo


Banedanmark must convert information from cable drawings into their GIS-system in order to deliver information in a fast, precise and effective way to people who are planning excavation work on Banedanmark property.

Vectorized cable information will give excavation workers better access to vital information and prevent damaging of Banedanmark’s cables. Consequently, this will result in fewer accidents and subsequently fewer delayed trains.

Besides securing improved access to digital asset data, vectorized cable information, made available in a GIS-system, will present new ways of using data across systems and processes in Banedanmark. This will be of great use for Banedanmark and cooperating partners.


Cable drawing

Cable drawings

Banedanmark currently stores cable drawings in formats such as TIFF, PDF, DWG and DGN.

The age and quality of the drawings are very diverse: the oldest drawings date back to 1920 and are most likely handwritten and scanned, whereas the newest drawings are developed and stored as CAD, 3D or BIM files.

Regardless of the format, the drawings must be converted into Banedanmark’s GIS-system. However, converting drawings saved as PDF or TIFF is the most complicated and substantial challenge as they account for 75 % of the drawings.

Many of the drawings are not geo-referenced, so Banedanmark is very interested in an automated solution for this.

However, the drawings contain large quantities of other information that might be helpful, for example information about the placement of the track, buildings in the area, geographical circumstances such as hills, bluffs etc.

Therefore, the automated solution must be able to identify vital cable information on the drawings such as information about the type and placing of the cable.

In some cases, the cable information can be identified from the title block of the document. In other cases, it can be identified from the legend of the document.

Banedanmark has uploaded examples of cable drawings that the participants of the challenge can use as inspiration. You will find them at the bottom of the page - see an example here.


Requirements and expectations regarding format and content of the solution

Technical requirements
Banedanmark is using ArcGIS from Esri as GIS-software platform. Therefore, the solution must be compatible with Esri’s file formats, which means that the digitalized GIS-data must be delivered in Esri’s file geodatabase format (.gdb) or in the standard format Shape (.shp).

Banedanmark currently runs on ArcGIS version 10.6.1 or ArcGIS Pro 2.1.0.

The GIS data projection should be delivered in the system of coordinates ETRS_1989_UTM_ZONE_32N.

Please be aware that the system of coordinates in the attached files can be in another or even unknown format.

Other information on the drawings, e.g. cable ID and cable specification, must be converted as well.

For inspiriration it can be helpful to visit Agency for Datasupply and Efficiency's webpage where it is possible to view historical maps for comparison, and the webpage might also be helpful for georeferencing the cable drawings.

The title block for the cable drawings is attached as well.

Solution requirements

Banedanmark wants to find a way to automate the converting process and any solution proposal should reflect that.

Therefore, it is important that suggested solution proposals can identify the relevant cable information from the drawings:

  • Type of cable
  • Geolocation of cable.

In addition, the solution proposal should also address how the gathered cable information is vectorized in GIS.  

The participants must be able to demonstrate a solution to the above requirements – either visually or theoretically.

Banedanmark will use the best solution proposals as background for a public tender.

Evaluation criteria

Following assessment requirements are used for basis of the evaluation:


  • To which degree does the solution fulfil Banedanmark’s needs?
  • To which degree is the solution automated?
  • Does the solution take Banedanmark’s technical requirements into account?


  • Is the solution practically possible?
  • Is the solution technically possible?
  • Is the solution possible to implement in real life?


  • Is the solution financially possible?


  • Is the solution innovative?
  • How good is the synergy between the assessment requirements?


The contestants with the most qualified solution proposals are invited to pitch their solution to the jury.

The jury selects the finalists based on the defined evaluation criteria.

The finalist participants can expect to receive questions to elaborate on when presenting to the jury on 22 January 2020.  

The presentation can take place via Skype or at an actual meeting.


A jury has been appointed to qualify and select the best solution proposals.

The jury consists of experts with different competences relevant to the subject.




Jury member

Jury member

Thor Gerner Nielsen

Radius Elnet /Dansk Ledningsejerforum


Janni Pia Petersen


GIS specialist

Glenn Bukh Andersen



Jury member

Jury member


Benjamin Biering


Head of Data Science

AI and ML expert

Mikael Munck


Ceo & Founder

Master in Economics & Finance.










Ideas for this challenge (20)
I vedhæftede offentlige fil er mit bud på hvordan opgaven kan deles op og understøttes med smarte it-løsninger.
Tegningerne er komplekse, forskelligartede og af varierende kvalitet. Det gør det til en stor opgave at udarbejde brugbare træningsdata til et ML-baseret system. Vi foreslår AI-baserede værktøjer, som kan automatisere en manuel registrering.
Løsningsforslaget er baseret på kendt teknologi, en kombination af automatiske og manuelle processer, med fokus på præcision og anvendelighed af det færdige digitale produkt, samt faseopdeling af projektet.
FME kan bruges til at trække ud ledningsoplysninger fra gamle tegninger. FME kan bruges til at læse og separere ud de geografiske objekter som findes i tegninger og gemme i et GIS format. FME kan integreres mod forskellige Machine Learning tjenester.
Få overblik over ledningsplaner ved at analysere dem efter kompleksitet
Ideen er at bruge det allerede eksisterende machine learning værktøj InfoExtract (udviklet af Atkins) til at træne en model til at identificere en geometrisk beskrivelse af ledningerne ud fra en kombination af en signaturforklaring og tegningshoveder
We propose a solution that starts as a semi-automated system and learns with each user interaction. Each digitized plan leads towards a more automated system until the human only has to validate propositions.
BaneApp kan nedbringe kabelskader og forsinkeler med denne data transformation der giver entreprenørerne hurtig og nemt tilgængelige tegninger. Organiseret og tilgængelig tegningsdata opfylder LER på kort og langt sigt.
This is a complex problem and would need several models to solve. Recent advances in deep learning may help.
Jeg er ikke specialist og derfor kan idéen virke helt tosset. I stedet for at gen-tegne kortene i et GIS program, kunne man ikke forsyne et tog med sådan en metaldetektor, som under kørslen kunne skanne alle "virkelige" ledninger under/omkring skinnerne direkte ind i et GIS program, hvorefter ledninger bliver navngivet ved hjælp af de gamle kort?
Idéen er at anvende FME Desktop og FME Server, dels til at udføre selve tolknings- og- konverterings-jobbet, og dels til at være det gennemgående bindeled i automatiseringsprocessen.
A Conscious Digitalisation Machine Learning Program
En perfekt kombination af manuel arbejdskraft (crowdsourcing) og en teknisk platform for nem og hurtig indtast af data til oversættelse.