In this paper, a fully automatic scheme for hiding digital watermarks into videoconference objects is proposed. Initially, a face and body detection module is implemented in order to provide an initial estimation of the foreground object. Next, each foreground object is decomposed using the Shape Adaptive Discrete Wavelet Transform (SA-DWT). Afterwards Object Adapted Qualified Significant Wavelet Trees (OA-QSWTs) are estimated for a pair of subbands, provided by the SA-DWT. Finally visually recognizable watermark patterns are redundantly embedded to the coefficients of the highest energy OA-QSWTs and the invert SA-DWT is applied to provide the watermarked videoconference object. Watermarked objects undergo various signal distortions and experimental results exhibit the efficiency and robustness of the proposed scheme.