Under specific inflow conditions, wind turbines in parked conditions with blades pitched to full feather (90°) can experience Stall-Induced Vibrations (SIV), leading to significant edgewise blade loads and potential blade failure. To enhance safety during standstill and enable control actions, detecting SIV is crucial. In this study, two detction methods are explored: (1) Monitoring of edgewise vibration peaks and (2) a Multi-Model (MM) state estimation approach. The peak monitoring method establishes baseline vibration peak levels and identifies deviations from these normal levels, offering a straightforward approach for early detection. The MM estimator method, in contrast, relies on a dynamic model of the vibrating blade that accounts for both lightly damped linear oscillations as well as non-linear SIV in which the vibrations grow and sustain into a limit cycle. The model, derived from a differential equation describing turbine blade vibrations, incorporates a forcing term parameterized by wind speed, yaw angle, and other parameters to switch between linear and nonlinear behaviors. The model is extended to a stochastic differential equation to include turbulent wind inflow. A Multi Model (MM) estimator based on the Interacting Multiple Model method is used to estimate states and parameters of the system. Once the parameters are estimated, SIV is detected based on the estimated coefficients. The models are evaluated on blade load data obtained from a measurement campaign on a standstill wind turbine. The results show that both the methods capture the presence of strong vibrations well. While simpler methods like the peak monitoring method might sometimes enable a quicker detection of SIV, they are prone to false alarms. Model based detection methods, like the MM estimator method might have a slightly delayed detection but can also provide information about the decreasing trend of vibrations, which can be useful in devising control strategies.