Content analysis technologies give more and more emphasis on multimedia semantics. However most watermarking systems are frame-oriented and do not focus on the protection of semantic regions. As a result, they fail to protect semantic content especially in case of the copy-paste attack. In this framework, a novel unsupervised semantic region watermark encoding scheme is proposed. The proposed scheme is applied to human objects, localized by a face and body detection method that is based on an adaptive two-dimensional Gaussian model of skin color distribution. Next, an invariant method is designed, based on Hu moments, for properly encoding the watermark information into each semantic region. Finally, experiments are carried out, illustrating the advantages of the proposed scheme, such as robustness to RST and copy-paste attacks, and low overhead transmission.