Some types of manufactured parts like sheet metals and skins often have a significantly different shape in a free-state position compared to their state-of-use position (as defined by their nominal CAD models) due to a combination of gravity and/or the residual effects of stress. Traditionally known as flexible (nonrigid compliant) parts, these dedicated fixtures are used for inspection operations in order to maintain flexible parts from a free-state position to a state-of-use position. This paper introduces a new automatic defect identification method primarily intended for two less-investigated manufacturing defect types: contour profile errors and hole localization. By combining simple techniques such as mesh boundary detection, fast boundary-based correspondence searches and accurate fast marching on triangulated meshes, the semi-geodesic distances from each boundary vertex on the acquired SCAN mesh to all the other boundary vertices is calculated, stored in a table and then compared to the corresponding values on the part's nominal CAD mesh. The comparisons found in the tables result in an estimation of the location and amplitude of the two aforementioned defect types. Compared to other work in this field, the overall approach does not rely on any mesh registration or finite-element analysis with tedious boundary conditions setup. It is also relatively fast. A fast algorithm/app based on this method was named the AFDA (Automatic Free-state Defect Approximation) and was validated against case studies in the aerospace sector. The results reflect the utility and effectiveness of the proposed approach.