Abstract—Given the substantial progress in advanced manufacturing, reaching the micro-scale has become necessary in several industries, such as the sensor, semiconductor, and aerospace sectors. This requires effective methods to detect, classify, and determine the topological parameters of micro-scale features on a workpiece. In this paper, microstructures were imprinted using the femtosecond laser machine. To accurately detect and classify these microstructures, Artificial Intelligence (AI) techniques were employed. This study integrated Convolutional Neural Networks (CNNs), specifically GoogleNet and ResNet-50, to achieve this goal. Various microstructures, also known as dimples, were simulated and used as proof of concept to train the CNNs and determine appropriate input parameters, which were subsequently used for training on the experimental dataset. With varying initial learning rates and mini-batch sizes, GoogleNet and ResNet-50 achieved validation accuracy typically ranging from 80% to 100% across simulated textures, with higher consistency observed in round and oval geometries. In the experimental dataset, accuracies generally ranged between about 80% and 90% for GoogleNet, while ResNet-50 exhibited wider variations, approximately between 50% and 95% depending on processing conditions. Both networks demonstrated strong performance in distinguishing textures produced under different marking speeds and layer conditions, effectively identifying low and high-gradient patterns. The results, thus, indicate that CNN-based approaches can reliably classify femtosecond laser-induced micro-textures, with ResNet-50 providing more stable performance across varying conditions and GoogleNet offering faster yet slightly more variable predictions. This multi-stage approach provides a comprehensive assessment of CNN-based micro-texture analysis and introduces an effective framework for quality control in manufacturing.

Keywords—femtosecond laser, convolutional neural networks, GoogleNet, ResNet-50

Cite: Cynthia AlLabaky and Roland Bejjani, "Convolutional Neural Network Classification for Femtosecond Laser Texturing," International Journal of Mechanical Engineering and Robotics Research, Vol. 15, No. 3, pp. 303-314, 2026. doi: 10.18178/ijmerr.15.3.303-314

Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).