Image Detection in Low-Quality Videos

Abstract

This article aims to describe recent findings on a prototype for assisting blind people. To improve its functioning the main approach is to build an intelligent system composed by Machine Learning of several models to detect and recognize multiple objects. The scope of this activity includes efficiency assessment, video data compilation, image segmentation, Data Mining processing, and tagging. This work also evaluates and depicts certain techniques and approaches to be applied to create models with high pattern detection efficiency. The algorithm must be light as well as quick, in order to be used in standard cell phones to assist blind people and provide meaningful information to the user as well as a small analysis of the results. Furthermore this study outlines specific methods and strategies employed to develop highly efficient models for pattern recognition.