We propose a modified Vicsek-like model to study influence dynamics and opinion formation in social networks. We work on the premise that opinions of members of a group may be considered to be analogous to the direction of motion of a particle in space. The opinions are susceptible to change under the influence of familiar individuals who maintain similar beliefs. This is unlike the bounded-confidence models which solely rely on interactions based on closeness of opinions. The influence network evolves either when similar-minded individuals acquaint or when they fall out over their beliefs. This yields an adaptive network to which are assigned dynamic centrality scores and varying influence strengths. A mix of individuals - rigid and flexible - is assumed to constitute groups - liberal and conservative. We analyse emergent group behaviours subject to different initial conditions, agent types, their densities and tolerances. The model accurately predicts the role of rigid agents in hampering consensus. Also, a few structural properties of the dynamic network, which result as a consequence of the proposed model have been established.