University of Twente Student Theses
Improved Low-contrast Spaghetti Defect Detection for FDM Printers
Leenheer, Bart (2025) Improved Low-contrast Spaghetti Defect Detection for FDM Printers.
|
PDF
33MB |
| Abstract: | Spaghetti defects are a common failure in FDM printing, often caused by object detachment from the print bed or missing supports, leading to filament being extruded in the air, These defects mostly cause unrecoverable errors, resulting in wasted time and materials. While computer vision-based anomaly detection methods exist, they lack robustness for low-contrast environments, for example, when using black filament against a dark background. This work explores whether a Low-Light Image Enhancement (LLIE) algorithm can improve anomaly detection in such conditions. A novel LLIE approach, CoLIE, is combined with a YOLO11 object detection model to perform anomaly detection, which is evaluated on a Colour, Black and external test dataset. The results are then compared to Obico, an open-source YOLOv2-based spaghetti detection model. To further analyze the practical performance, a new metric is introduced, the Print Failure Stopping Metric (PFSM), which highlights the performance per print. Results show that CoLIE does not consistently improve the detection performance and even reduces it in some cases. However, YOLO11 shows a significant improvement over Obico, even achieving a 17.5% higher F1-score on the external test dataset, indicating good generalizability. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science MSc (60300) |
| Link to this item: | https://purl.utwente.nl/essays/106149 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page