RGBD Video Automatic Annotation for Training Segmentation Models to Guide Blind People

Institut

Professur für autonome Fahrzeugsysteme

Typ

Semesterarbeit / Masterarbeit /

Inhalt

experimentell / konstruktiv /  

Beschreibung

Also possible as IDP, research internship (Forschungspraxis), guided research, etc.  

This research aims to develop an automatic annotation system for RGBD video data to facilitate the training of segmentation models for guiding blind people. The project involves creating an annotation pipeline that generates precise masks for recorded video frames, leveraging both RGB and depth information. This automated process will significantly reduce the manual effort and time required for data labeling, enabling the creation of large annotated datasets essential for training robust segmentation algorithms. The effectiveness of the annotated data will be evaluated by training segmentation models and testing their performance in real-world navigation tasks for blind users.

This research is in collaboration with the Startup SmartAIs (https://de.linkedin.com/company/smartais)

Voraussetzungen

• Coding experience (Python) 
• Foundational knowledge in 2D-3D computer vision 
• Curious mindset 
• Independent and organized work attitude 

Möglicher Beginn

sofort

Kontakt

Yamen Mohisn
yamen.mohisnsmartais.de