A supervised Master’s thesis developed a mobile prototype for inventorying civil-protection facilities for the Swiss Federal Office for Civil Protection. An adapted neural network recognises equipment such as air filters, protective doors and fire extinguishers directly on a smartphone. Local processing enables use without a connection to an external server. Adrian F. Meyer provided technical supervision and supported the development and assessment of the object-detection methods.

Context and significance

Mobile computer vision can assist on-site inventory work without necessarily uploading photographs to a cloud service. The prototype explores this for defined equipment categories in civil-protection facilities. Practical usefulness also depends on robust recognition under changing light, viewing angles and device constraints.

01

The research question

How can equipment be recognised directly on a smartphone?

02

Finding an answer

  1. Develop a mobile prototype within a supervised Master’s thesis.
  2. Adapt a neural network to equipment such as air filters, protective doors and fire extinguishers.
  3. Assess local object detection without a connection to an external server.

03

Findings and applications

  • A mobile prototype for inventorying civil-protection facilities.
  • Technical supervision of the development and evaluation of object-detection methods.