I have read some research papers (https://ieeexplore.ieee.org/document/7362651 , ) wherein they added changes to OpenCV functions (DetectMultiscale, DetectSinglescale) to achieve execution time speedup. How to achieve this? Help will be highly appreciable.