Action recognition on video data using pose estimation

Context

Correctional facilities (JVA) are obliged to look after the health of prisoners. In the day-to-day work of the prison authorities, critical situations repeatedly arise between inmates that require rapid recognition and intervention by the prison staff in charge.
At the FZI, you will support us in the use of state-of-the-art camera technology and artificial intelligence for action and situation recognition, as well as the behavioral analysis of people, in order to significantly support prison staff in recognizing critical situations and to facilitate the responsible work of preventing violence and ensuring security in the prison.

Tasks

• You will familiarize yourself with the existing framework for action recognition.
• You will conduct research on suitable AI models for action and situation recognition as well as behavior analysis.
• You will design and implement new AI models for action and situation recognition as well as behavior analysis.
• You compare different AI models and analyze how they can be adapted to the respective situation.
• You will train the AI models and evaluate the system with test subjects.

Prerequisites

• You are studying electrical engineering and information technology or a related course.
• You are interested in the topic and have some experience with artificial intelligence and computer vision.
• You have good programming skills and experience with the Python programming language and AI frameworks (PyTorch, TensorFlow).
• You enjoy working independently, are motivated and committed.
• You have very good written and spoken German and English skills.