Performance Optimization of Object Tracking Algorithms in OpenCV on GPUs

Machine-learning-based computer vision is increasingly versatile and being leveraged by a wide range of smart devices. Due to the limited performance/energy budget computing units in devices, careful implementation algorithms critical. In this paper, we analyze performance bottleneck two well-known for object tracking: detection optical flow Open-source Computer Vision library (OpenCV). Based on our in-depth analysis their implementation, found current fails utilize Open Computing Language (OpenCL) accelerators (e.g., GPUs). analysis, propose several optimization strategies apply them OpenCL tracking algorithms. Our evaluation results demonstrate improved up 86% 10%. We believe can be applied other implemented OpenCL.