This paper introduces a distributed observer-based protocol to achieve bipartite leader-follower consensus for linear multi-agent systems under intermittent failures. Under possible failures in the measurement and control input channels, the consensus protocol is based on intermittent interaction rules. Using the concept of time scales theory, the closed-loop system dynamics are converted to a switched system between continuous-time and discrete-time dynamics. Closed-loop stability is proved using Lyapunov functions and time scales theory. Numerical simulations illustrate the effectiveness of the designed distributed observer-based protocol proposed in this paper.