Visual Inspection for Industrial Quality Assurance

The thesis project focuses on the development of an artificial intelligence-based visual inspection system for quality control of a product in the manufacturing industry. More specifically, several modules and components of an entire pipeline, that receives the product image as input and returns a result concerning the conformity of the product from an aesthetic point of view, were implemented. The main aspects checked are the integrity of the product (i.e., presence of scratches, dents, grease stains), and the readability and correctness of the print on the product. For this project, techniques of object detection, anomaly detection for computer vision, and optical character recognition were utilised.