A smart camera processor has to perform substantial amount of processing of data-intensive operations. Hence, it is vital to identify critical segments of the processing load by involving HW/SW codesign in smart camera system design. This paper presents a novel fully automatic hybrid framework that combines heuristic and knowledge-based approaches to partition, allocate and schedule IP modules efficiently. In this work, the concept of Bayesian Belief Network (BBN) is utilised and incorporated into the proposed framework. In the experiment section of this paper, we report a comparison of our proposed framework with three previously published work: A BBN based method proposed by a research group from the University of Arizona, the exhaustive algorithm and finally the with greedy algorithms.