Design of an efficient ASIP-based processor for object detection using AdaBoost algorithm

Object detection is at the heart of nearly all the computer vision systems. Standard off-the-shelf embedded processors are hard to meet the trade-offs among performance, power consumption and flexibility for future algorithms required by object detection applications. Therefore, this paper presents an Application Specific Instruction set Processor (ASIP) for object detection by using AdaBoost-based learning algorithm with Haar-like features as weak classifiers. In the proposed ASIP, Single Instruction Multiple Data (SIMD) architecture is adopted for fully exploiting data-level parallelism inherent to the target algorithm. With the custom instructions and application-specific registers, the highest computational task within AdaBoost algorithm, which takes almost 80% of the total cycle counts, is accelerated by a factor of 14x compared to the baseline processor.