Object detection is an important function for intelligent multimedia processing, but its computational complexity prevented its pervasive uses in consumer electronics. Cost-effective & energy-efficient computations can be provided with various innovative multicore architectures. However, extensive optimizations are needed to unravel inherent parallelism of multicore processing. This paper presents interleaved task reordering and task splitting of parallel tasks in object detection for multicore platform. A simple face detection prototype has been implemented on Sony PS3 with Cell Broadband Engine, where our proposed two methods improve the overall performance by 10% & 19% respectively.