Context
Automatic defect detection in tomographic volumes using artificial intelligence approaches

For the foundry industry, which produces metal parts with "tormented" shapes, often with internal ducts, tomography is a very interesting tool for analyzing the internal health of parts. Discontinuities in thickness can be positioned and visualized in several orientations. This makes it possible to consider part acceptance criteria based on the position of the discontinuity in relation to the surface, or in relation to a highly stressed critical zone.

At present, conformity of the internal health of castings is achieved by comparing 2D X-ray images with reference images available from ASTM