The noldus database: Automated recognition of restaurant related activities for the restaurant of the future

Moritz Kaiser, Dejan Arsić, Benedikt Hörnler, Martin Hofmann, Gerhard Rigoll

Research output: Contribution to journalConference articlepeer-review

2 Scopus citations

Abstract

The'Restaurant of the Future' is a futuristic approach for consumer research in food-related scenarios. Customers are closely monitored by cameras. Currently the video footage is evaluated and annotated manually. For a gradual automation of this process a video database with eating scenarios is published. The database is quite challenging with light changes, overexposure, occluded objects, and equally colored objects, such as glass, plate, skin, and cloth. Basic activities, as drinking, eating, coming in, going out, etc. are manually annotated and can be used as ground truth. A baseline system, consisting of five recognition modules and a rule-based activity recognition method is presented in this treatise. The results can be used as baseline results for comparison within the community.

Original languageEnglish
JournalInternational Workshop on Image Analysis for Multimedia Interactive Services
StatePublished - 2011
Event12th International Workshop on Image Analysis for Multimedia Interactive Services, WIAMIS 2011 - Delft, Netherlands
Duration: 13 Apr 201115 Apr 2011

Keywords

  • Behavior recognition
  • Multimedia databases
  • Object recognition
  • Restaurant of the futurec
  • Tracking

Fingerprint

Dive into the research topics of 'The noldus database: Automated recognition of restaurant related activities for the restaurant of the future'. Together they form a unique fingerprint.

Cite this