Bachelor-/ Master Thesis »Synthetic Data for Deep learning Models in production«

The Fraunhofer-Gesellschaft (www.fraunhofer.com) currently operates 76 institutes and research institutions throughout Germany and is the world’s leading applied research organization. Around 30 800 employees work with an annual research budget of 3.0 billion euros.

At the Fraunhofer IPT in Aachen, we work with more than 530 employees every day to make the production of the future more digital, more flexible and more sustainable. In the department »Production Quality«, we deal with the digitalization of production systems in order to increase quality, resilience and sustainability in production.

The application of deep learning models to automate visual process monitoring and quality control enables quality and efficiency gains in production. In this context, a typical challenge is the lack of representative training data, such as data of quality-critical anomalies. To increase the performance of deep learning models in these applications, customized pipelines for image augmentation must be designed that enrich existing datasets with synthetic images. Designing effective augmentation pipelines requires domain knowledge and is so far mainly done manually. Within the scope of your thesis, you will investigate how optimization methods (e. g., Bayesian optimization) can be used to optimize augmentation pipelines for individual use cases in production. The main potential lies in the targeted combination of different augmentation types (e. g. geometric, color and mixing-based transformations) and the integration of domain knowledge.

What you will do

Literature research on the topics of deep learning, image augmentation and global optimization methods
Identification of requirements for the design of efficient augmentation pipelines in production use cases
Development and implementation of a method for optimizing augmentation pipelines
Experimental validation based on a practical use case in the field of computer vision (e. g., visual quality control)
Preparation and documentation of results

What you bring to the table

You are studying mechanical engineering, industrial engineering, computer science or a comparable subject
You have first experience in Python
You have basic knowledge of the theory and methods in machine/deep learning
A self-responsible and structured way of working
Good language skills in German and/or English

What you can expect

Scientific work on a current and practice-relevant topic
Professional supervision and support for your thesis
Participation in innovative research and industry projects with industry partners
A state-of-the-art machine park equipped with edge cloud systems and 5G infrastructure

We value and promote the diversity of our employees' skills and therefore welcome all applications - regardless of age, gender, nationality, ethnic and social origin, religion, ideology, disability, sexual orientation and identity. Severely disabled persons are given preference in the event of equal suitability.

With its focus on developing key technologies that are vital for the future and enabling the commercial utilization of this work by business and industry, Fraunhofer plays a central role in the innovation process. As a pioneer and catalyst for groundbreaking developments and scientific excellence, Fraunhofer helps shape society now and in the future.

Interested? Apply online now. We look forward to getting to know you!

Maximilian Motz M.Sc.
Research assistant »Production Quality«
Phone: +49 241 8904-449

Fraunhofer Institute for Production Technology IPT

www.ipt.fraunhofer.de

Requisition Number: 69829 Application Deadline:

Job Segment: QC, Database, Training, Research Assistant, Quality, Education, Technology, Research

Apply now »