Thesis Proposal: AI-Based Detection and Classification of Inclusions in Metals

We are looking for a Master's student to join our materials engineering and AI team for a challenging graduation project: AI-Based Detection and Classification of Inclusions in Metals. In this thesis, you will develop an intelligent model that supports metallurgical engineers in assessing metal cleanliness more efficiently and objectively.

Pros of this thesis:

• Apply cutting-edge AI in a real-world industrial and research context
• Collaborate closely with metallurgical engineers and laboratory professionals
• Gain experience with practical data acquisition, AI model development, and performance validation
• Contribute to the automation and standardization of metallographic analysis

What you'll be doing

The goal of this thesis is to develop an AI-based model capable of automatically identifying, classifying, and quantifying inclusions in metallic samples. The model should determine:

• The type of inclusion (A, B, C, D, or E-type, according to standard metallographic classification systems such as ASTM E45 or DIN 50602),
• The magnitude and size distribution , and overall cleanliness rating of a given sample.

This system will assist metallurgical engineers and laboratory personnel by providing faster, more consistent, and data-driven inclusion analysis.

Scope and Methodology

• Conduct a literature and standards review on metallurgical inclusion classification (ASTM E45, ISO 4967, DIN 50602) and existing AI applications in materials inspection
• Analyze current laboratory workflows, including manual inspection methods and microscope software capabilities
• Collect and prepare data by acquiring and annotating microscope images of metal samples with various inclusion types and cleanliness levels
• Develop and train an AI model (e.g., CNN or transformer-based vision model) to automatically detect, classify, and quantify inclusions
• Validate and evaluate model results against expert laboratory assessments, focusing on accuracy, precision, and consistency
• Propose integration options for the developed model into existing laboratory workflows or microscope software, and present a prototype demonstrator

Expected outcomes

• A trained and validated AI model for inclusion detection and classification
• Quantitative performance comparison between AI and manual inspection results
• Recommendations for implementing automated cleanliness evaluation in laboratory environments
• Final thesis report and presentation

Your profile

• Master's student in one of the following fields: Artificial Intelligence, Data Science, Applied Computer Science, Computational Materials Science, Mechanical or Materials Engineering (with strong data/AI interest)
• Programming experience in Python (TensorFlow, PyTorch, OpenCV)
• Basic understanding of machine learning and image processing
• Analytical mindset and interest in applying AI to engineering and manufacturing challenges
• Experience with model validation and statistical analysis is a plus