Identification of brand logos from video feed

Abstract

In this paper we have made an attempt to build a Machine Learning model based on Yolo algorithm to calculate the brand visibility of sponsored advertisements during live sports events such as cricket matches and based on brand visibility, we have calculated effective and accurate advertising cost. To calculate the brand visibility, we have restricted ourselves to brand logos only which are visible on cameras. As a result of our analysis, we observed that in a cricket match Brand A had 26% visibility while brand B had 67% visibility and yet brand A was charged much higher than brand B by the advertising companies and hence machine learning based analytical tools can be of great use while effectively calculating the advertising costs.