<p>Ensuring the surface quality of industrial components requires the detection of small superficial defects—such as cracks, bumps, and peaks—using high-resolution 3D sensors. However, training machine learning algorithms for this task is constrained by the limited availability of annotated 3D defect datasets. In this work, we propose a method for generating synthetic 3D datasets of surface defects using Free-Form Deformation applied to triangular meshes. The technique allows localized insertion of parametrized defects adapted to the object’s geometry and supports diverse defect types through customizable elevation maps. We validate our approach by providing a direct comparison between synthetic defects and real scanned surfaces, demonstrating that the generated defects capture key geometric features and variations. Also, we present a case study in which a synthetic defect dataset—generated by simulating the acquisition of a profilometric 3D sensor, including surface and sensor noise—is used to train object detection networks, which are then evaluated on real scans. The results demonstrate that the synthetic defects effectively reproduce the essential characteristics of real defects, providing a scalable and versatile tool for both dataset generation and the development of inspection systems in industrial contexts.</p>