Research Seminar on AI: Deep Anomaly Detection for Automated Visual Inspection

 

Dienstag, 28.06.2022, 16.00 Uhr

The Global Visual Inspection Systems market is expected to grow at a CAGR of around 13.5% from 2020 to 2027, reaching a market value of over US$ 25.5 Bn by 2027. One of the main drivers behind this projected growth is AI, which has the potential to facilitate the automated inspection of ever more complex products, which was previously thought to be impossible. To realize this potential, however, AI needs to better fit the requirements of the Visual Inspection industry. Specifically, manually collecting and labeling a large number of defective as well as defect-free products, as required by current supervised learning schemes, is infeasible.
In this talk, Oliver Rippel will give an overview of Anomaly Detection, a technique free from the limitations inherent to supervised learning. He will first show how Visual Inspection can be cast as an Anomaly Detection problem. Afterwards, he will detail the algorithmic advances in recent years fueled by the advent of Deep Learning. He will conclude the talk by identifying trends and giving recommendations to practitioners in the field.

Oliver Rippel is a final-year PhD student with the Institute of Imaging & Computer Vision at RWTH Aachen University, supervised by Prof. Dorit Merhof. His research interests include Anomaly Detection and Out-of-Distribution detection in context of automated visual inspection. Additionally, he is also passionate about translating research insights into industrial applications.