Nowadays, the real-time seeing estimation at the focus of a telescope is strongly emphasized because this knowledge virtually drives the dimensioning of adaptive optics systems and the operational aspects of instruments. In this context, we study the importance of using images from an active optics Shack-Hartmann (AOSH) sensor to provide accurate estimates of the seeing. At the critical location of the focus of the telescope, the AOSH sensor delivers long-exposure spot point spread functions, which are directly related to the atmospheric seeing in the line of sight. Although AOSH sensors are specified to measure not spot sizes but slopes, we show that an accurate estimation of the seeing can be obtained from AOSH images using a dedicated algorithm. We analyse the sensitivity of two algorithms and compare them to various parameters in a systematic way, demonstrating that an efficient estimation of the seeing can be obtained by adequate means.