Self-sensing allows to use a Dielectric Elastomer Actuator (DEA) simultaneously as an actuator and sensor, without the need of external sensors. DEAs are composed of a dielectric elastomer that is sandwiched between two electrodes. If a voltage difference is applied to the electrodes, the DEA will deform, resulting in an increase in surface area and a decrease in thickness. Different self-sensing techniques are suggested in the literature to estimate the state of a DEA. Among other self-sensing techniques an Extended Kalman Filter (EKF) can be used for self-sensing, to achieve accurate estimations of the DEA Deformation. However, the EKF needs to be correctly calibrated to achieve precise and stable estimations. The results of this thesis shows the implementation of four different EKF models and their calibration. The calibration of the EKFs is implemented using an optimization algorithm. The algorithm uses the Root Mean Square Error (RMSE) between the estimated planar stretch ratio of the DEA and the stretch ratio detected using a camera together with Computer Vision. The advantage of using an EKF for self-sensing is that the implementation does not depend on the shape nor size of the DEA and that no superposed driving signal is needed to actuate the DEA. Thus the EKF implementation proposed in this thesis is a versatile self-sensing technique for DEAs.