Abstract
This study investigates the application of deep learning-based image segmentation using 2D optical imaging for the microstructural characterisation of composite materials with hybridised fibres, offering a potentially cost-effective alternative to computed tomography/3D imaging. Laminates were produced using the HiPerDiF method, combining discontinuous carbon and basalt fibres to reinforce a poly(L-lactic acid) (PLA) matrix. The results demonstrate that the Generalised Dice Loss function significantly outperforms others, particularly in the void class, achieving a 19% improvement in Dice Similarity Score on an unseen dataset for full image characterisation. Similarly, for Boundary Intersection over Union (IoU), which measures the accuracy of local boundary detail capture, the model trained with Generalised Dice Loss achieved 61.4%, compared to 55.0% for the next best model trained with Compound Loss. These findings suggest that regional loss functions are better suited for image-based microstructural characterisation and quality inspection. Additionally, volume fraction, relative fibre and void ratios, and fibre alignment computed from the segmentation results closely match ground truth data. Challenges related to data limitations and variability are also briefly discussed.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.