In recent years, the concept of the digital twin (DT) has gained substantial attention in the field of structural health monitoring. DTs have shown the potential to mirror the behaviour of physical structures in real-time, enabling a more comprehensive and timely understanding of structural performance. For civil and mechanical structures, finite element models (FEMs) are often preferred as representations of DTs due to their capability for mechanical analysis, particularly in analyzing responses to different loads. Generally, structures may encounter crack damage, and currently, there is no widely proven approach to twin or incorporate crack damage in FEM-based digital twin systems. Therefore, this paper proposes a crack twinning method for shell structures using computer vision and finite element method. Computer vision techniques, such as perspective transformation, image segmentation, and object tracking, are used to identify and track the propagation of cracks. The detected crack is then modeled and incorporated into the FEM, based on the phase field modelling method This proposed approach aims to continuously synchronize crack propagation in the FEM, making the twinning procedure interactive. The concept is validated through a simulation experiment on a plate structure. Finally, the effectiveness of the proposed method is discussed, highlighting its potential in advancing digital twin technology for structural health monitoring.