Project : AI Development for Visual Defect Detection on Multi-Color Knitted Textiles
Department : Co-supervision between the DEMCO Group and a Startup (PFE Book Link)
Context :
- The project is co-supervised by a team leader from the DEMCO Group.
- Working on-site at ViZmerald and DEMCO, this internship offers a unique blend of a startup’s agility and an industrial leader’s reputation, providing a dual learning opportunity.
Objectives :
- Extend the current AI solution to detect defects in knitted fabric types.
- Robustly handle color variations during production and report both color shade defects and shape inconsistencies.
Tasks :
- Actively participate in building the database for targeted defect detection, performing data cleaning and labeling in collaboration with the project team.
- Test ViZmerald’s current algorithm deployed on Nvidia Jetson-based Edge devices for textile defect detection and extend it using the updated database.
- Utilize auto-encoder techniques for anomaly detection and optimize its performance for automated deployment on ViZmerald’s ViAFaM machine.
- Enhance real-time execution results of the algorithm by improving ViZmerald’s current Web and Desktop interfaces to display performance dashboards tailored for the factory’s QHSE managers.
Supervisor : Team Leader from DEMCO and the ViZmerald Startup