A joint research team from Milan Polytechnic and Siemens has proposed a low-cost, robust in-situ detection framework for geometric deviations in Laser Powder Bed Fusion (L-PBF) lattice structures. By systematically calibrating image segmentation algorithms, the study successfully addresses detection challenges caused by variations in lighting, part size, and geometry. Experiments on spiral tetrakaihedron lattices of different sizes showed that after introducing controlled defects, the team could detect defects with a false positive rate below 2%. This technology offers a new industrial solution for improving the yield of complex metal additive manufacturing parts and reducing production costs.