Many offshore geotechnical problems--such as cyclic T-bar penetration and lateral buckling of pipelines--are affected by the tendency of fine-grained soils to strain-soften due to remoulding. Careful calibration of the constitutive model parameters that control strain softening is essential for accurate simulation of these processes. This is usually achieved by matching constitutive model response with standard element test data, which implicitly assumes that homogeneous stress/strain fields exist within the sample. However, popular element test protocols (e.g., triaxial, simple shear) cannot apply sufficient deformation--at least as measured at the boundaries--to achieve fully remoulded conditions. This work explores the potential to determine strain-softening parameters experimentally, directly from image-based full-field deformation measurements and external loading data. Artificial data generated from finite element simulations of biaxial compression and T-bar penetration tests, using a non-locally regularised strain-softening constitutive model based on Modified Cam Clay, are then used to demonstrate the potential of the proposed technique. The results demonstrate significant potential for the application of the technique to identify constitutive parameters from full-field measurements even when polluted with modest measurement noise.