Anomaly detection in videos

  • Subject:Anomaly Detection, KI, Computer Vision, Handlungserkennung, Human Action Recognition
  • Type:Bachelor- / Masterarbeit
  • Date:ab 12 / 2024
  • Tutor:

    M. Sc. Dominik Beyer

  • Zusatzfeld:

    Thesis at the FZI.

Anomaly detection in videos

Context

The demand for advanced AI solutions for anomaly detection in videos is increasing. You will work on novel technologies that are used in challenging scenarios, especially where it is difficult to obtain suitable data material for the respective situation. Among other things, your work will help to detect critical activities in real time, improve security and create innovative systems with real added value.

Tasks

  • Researching suitable AI models for anomaly detection and behavior analysis
  • Selection and implementation of suitable AI models for anomaly detection.
  • Comparison of different AI models and analysis of how they can be adapted to the respective situation
  • Design of an evaluation concept and definition of anomaly scenarios
  • Training of the AI models and evaluation of the system at the FZI with test persons

Prerequisites

  • Motivation and interest in independently solving technical problems
  • Knowledge of machine learning and artificial intelligence
  • Experience with Python and AI frameworks (e.g. PyTorch)
  • Analytical, problem-solving and communication skills