Initial Situation

Banedanmark needs to digitize more than 10,000 cable diagrams. This information must be available quickly, particularly during construction work, so that existing lines can be reliably taken into account and damage avoided. The cable diagrams vary greatly in age, quality, and scale, and their information density is enormous. Purely manual digitization would take several years.

Project Overview

Client
Banedanmark (Danish State Railways)

Sector
Public Transport

Solution
Automated digitization of thousands of cable diagrams.

System Environment
Over 10,000 cable diagrams of varying age, quality, and scale

Approach
Gradually increasing automation through human-machine collaboration

Award
Ranked in Banedanmark’s international competition for ideas

The SCS Contribution

SCS developed an automation concept designed to gradually reduce manual work steps. Human corrections provide new training data, enabling learning systems to take over an increasing number of tasks.

Learning with Every Correction

Human and machine work closely together in this concept. Manually verified and corrected results improve the database, allowing further work steps to be automated gradually.

Ground Truth Built Efficiently

Learning systems require reliably labeled training data. The SCS concept reduces the effort required by systematically combining existing diagrams with already known information.

Historical Diagrams Combined

Diagrams from different decades often exist for the same location. The concept localizes and overlays these documents. Known information in one diagram can thus help assign previously unknown details in another, similar to the principle of the Rosetta Stone.

Diagrams Localized Step by Step

First, character recognition is used to roughly determine which station or track section a diagram belongs to. The concept then provides for fine localization, for example by matching street structures with map data. Based on this, relevant objects can be recognized and additional object types can be trained.

Result

The SCS concept won Banedanmark’s international competition for ideas and expanded the technological solution space for the digitization of more than 10,000 cable diagrams. According to SCS, the developed approaches are being integrated into the actual project implementation. Instead of recording every historical diagram completely by hand, the concept combines multiple information sources and human corrections with learning processes. This should allow more and more steps of the digitization process to be automated.

Over 10,000 diagrams

to be digitally accessible at Banedanmark.

1st Place

in the international competition for ideas.

Human Plus Machine

Corrections gradually improve the database.

Services Utilized

Algorithms, ML & AI

Computer Vision

Public Transport

The bald man in the white shirt, smiling at the camera, represents the friendly SCS team.

Your Contact Person

Pascal Iselin

Algorithms, Analytics & AI
+41 43 456 16 00

For questions regarding machine learning, document analysis, and the automation of data-intensive digitization tasks.

Further Case Studies

  • Optischer Versuchsaufbau Vision

    Georg Fischer

    Vision for Groundbreaking Weld Seam Inspection

    Image-based inspection of weld seams.

  • White lab device labeled "zenCELL owl" utilizing edge computing; blurred lab utensils in the background.

    Manufacturer of zenCELL owl

    AI at the Edge – Counting Cells

    Deep-learning cell counting directly on the lab PC.