📄 中文摘要
硬涂层在工业中发挥着关键作用，陶瓷材料因其出色的硬度和热稳定性而广泛应用于需要优越机械性能的领域。然而，利用人工智能进行表面粗糙度分类时，通常受到对大量标记数据集和昂贵高分辨率成像设备需求的限制。研究提出使用通过Stable Diffusion XL生成的合成图像，作为实验数据的有效替代或补充，以实现陶瓷表面粗糙度的分类。研究结果表明，使用生成图像增强真实数据集后，测试准确率与仅使用实验图像获得的结果相当，证明了合成图像在这一领域的有效性。
评估合成图像作为表面粗糙度分类实验数据的有效替代品
📄 English Summary
Evaluating Synthetic Images as Effective Substitutes for Experimental Data in Surface Roughness Classification
Hard coatings are crucial in various industrial applications, with ceramic materials providing exceptional hardness and thermal stability for demanding mechanical performance. However, the deployment of artificial intelligence (AI) for surface roughness classification is often limited by the necessity for large labeled datasets and expensive high-resolution imaging equipment. This study explores the use of synthetic images generated by Stable Diffusion XL as an efficient alternative or supplement to experimentally acquired data for classifying ceramic surface roughness. Results demonstrate that augmenting authentic datasets with generative images achieves test accuracies comparable to those obtained using solely experimental images, validating the effectiveness of synthetic images in this context.