Improving the Accuracy of Detecting Objects on the Images Using the Haar Cascade, Neural Network and Parallel Programming

The paper discusses the possibility of increasing the accuracy of identifying static and dynamic objects on the image using the Haar cascade by carrying out additional filtering of detected objects using a multilayer deep learning class neural network. Separately, the possibility of performing parallel processing of one image by several Haar cascades on high-performance multicore or cluster systems for increase the accuracy of correctly identifying objects of interest on the image is considered.