This repository serves code for the project aimed at detection and locomotion tracking of customers at supermarket, and also collect relevant data such as approximate age group, total shopping time, etc. This data can then be utilised to make decisions for product placement in the brick and mortar store based on real trends.
At the end of the day, there are a few goals which must be met to ensure good sales at any supermarket. The prime goal of supermarkets for maximizing is not to give you the products you want, it is to make you buy things you didn't wish to. This also includes newly launched products with a low PR value and credibilty.
According to my hypothesis, sales of such products can be maximised by placing them in the most crowded sections of the crowd, where people shop more, spend more time, and hence these products have higher visibility. Also, it is psychologically proven that placing a mediocre or low credible products near famous products can make people want to buy them as well.
The problem however is to accurately find these areas. This can be done via detecting people and their shopping times, locomotion trajectory in the market. Using this real data, effective decisions can be made to revamp the sorting of products.