Master Thesis: Synthetic Data Generation for Defect Detection

Spring 2026, 30 hp

With headquarters and production in Solna, Rolling Optics AB is a pioneer in advanced micro-optics, creating security-grade visual effects that protect physical security documents and luxury products from counterfeiting. Our unique technology enables us to design and manufacture intricate optical patterns and structures — from concept to finished roll — all under one roof.

Context

Are you interested in writing your master's thesis with us at Rolling Optics? You are welcome to submit your application Once we receive your application, we'll review it and contact you for a follow-up discussion. During this dialogue, we'll have the opportunity to refine the topic together and align it with your interests and academic requirements.

Thesis Subject

As the demand for high-quality micro-optical materials grows, automated quality control becomes increasingly important. Detecting defects such as scratches, dirt, printing errors, and structural irregularities requires advanced algorithms, often involving AI and machine learning for computer vision. However, successful training of such models depends on the availability of large, high-quality datasets, something that can be challenging to obtain in practice.

Objective

Develop a method for generating synthetic images of micro-optical labels with simulated defects and validate that this data can be used to train effective defect detection models.

Example Tasks

* Investigate and document common defect types found in micro-optical materials

* Implement procedural generation of various defect types (e.g. scratches, dirt, print errors, structural defects) with adjustable parameters

* Develop 3D models and materials in Blender that replicate micro-optical structures and moiré effects

* Create an automated pipeline for large-scale dataset generation, varying lighting conditions, camera angles, and defect parameters

(Train and evaluate defect detection models (e.g. YOLO) using synthetic data, real-world data, and combined datasets

Profile / Skills

We're looking for a student who is curious, analytical, and eager to explore the intersection of computer graphics, computer vision, and manufacturing quality control. You work independently and systematically but also enjoy collaborating and discussing technical challenges with others.

Meriting Skills

* Experience with Blender and procedural generation

* Proficiency in Python programming

* Basic understanding of computer vision or machine learning

* Experience with the Blender Python API

* Knowledge of physically based rendering (PBR) and material simulation

* Understanding of optics and light scattering

Qualifications

* You are currently enrolled at a university and eligible to write your thesis during Spring 2026

* Your field of study is relevant to the topic (e.g. computer science, engineering physics, applied mathematics, or similar)

What We Offer

At Rolling Optics, you'll be part of an innovative environment where science meets design and precision manufacturing. We work at the frontier of micro-optics and are passionate about creating secure and sustainable solutions that protect authenticity and trust worldwide.

You'll have access to equipment, expertise, and a workspace at our Solna office, and you'll collaborate closely with our engineers and researchers. We believe in open knowledge-sharing and cross-disciplinary innovation, ensuring that your thesis contributes both to your academic goals and to real-world applications.

Because of the high-security nature of the work we do, we will sadly not be able to accommodate international students from non-EU countries. We thank potential applicants for their understanding.

Contact

Applications are reviewed on a rolling basis — we welcome your application today