Camera based in situ layer segmentation and object reconstruction for digital twins in FFF 3D printing

Reliability and quality assurance remain challenging tasks in Fused Filament Fabrication (FFF) 3D printing. A first step toward this goal is to generate a high-resolution reconstruction of the printed object. This paper presents a camera-based in situ method using low-cost hardware to create such a digital twin. The method captures high-resolution images of each printed layer immediately after the layer is finished. A U-Net++-based neural segmentation network is trained to isolate the freshly deposited material in the captured images. The segmentation attains an accuracy of 0.94 (intersection over union). Experiments show that the network generalizes well to unseen objects and materials. The high-resolution 3D object is reconstructed by extruding each segmented layer to its specified height and combining all layers into one model with a mean dimensional offset of 58.2 µm compared to the physical object. The generated model preserves internal structures and fine features, and can be used for quality assurance, geometric accuracy assessment, stress analysis, or as a basis for certification in safety-critical applications.