To accurately predict the ductile rupture or failure using uncoupled rupture models for a given material, specific model parameters are required. These parameters are difficult to determine in a direct approach from experiments and need to be estimated using a hybrid experimental and numerical analysis, which accuracy relies on the quality of the anisotropy and hardening model calibration. In this study, material model parameters are estimated for AA6016-T4 and AA5182-O thin sheets. The methodology to determine material parameters of a combination of Swift-Voce hardening law and Yld2004-18p yield criterion is based on inverse identification over a full database made of quasi-homogeneous tests and specific rupture tests. The experimental data are obtained from sheet metal samples in the form of either stress-strain curves or load-displacement curves and local strain evolution measured by digital image correlation. To validate the simulation results, three additional tests on notched specimen are considered. The failure model parameters for a shear modified uncoupled Lou’s rupture criterion are then determined using an average value of the triaxiality ratio and the Lode parameter at the material point of maximum equivalent plastic strain. The final aim of this study is the numerical prediction of the strength of a clinched joint of dissimilar AA6016-T4/AA5182-O sheets and the occurrence of rupture is numerically investigated at different stages.