DefinePK

Abstract

Retail automation is a comparably fashionable industry, particularly with the support of AI, ML, and computer vision as some of the facilitating forces. The technologies can assist the rise of customer satisfaction level and the business activity as they assist in identifying the different kinds of interactions between people and objects that can be purchased in the retail, thus providing a real-time view of the customers, their preferences, purchasing habits etc. The purpose of the paper is to speak about the use of automation, which can be offered in the event of human-object interaction detection, and its worth to the various facets of the retail process, such as the storage and systematization of stock and communication with customers. The classical problems and retail systems include Target identity, Product identity and occlusion besides limited real-time learning. However, the issue of distinguishing between human and object interaction using the AI models like CNNs eliminates these issues. New model defines and classifies the human activities and material things whenever carrying out customer and product-related activities; it improves productivity of the organization. Accuracy, precision, recall, F1-score and map evaluation measures with the assistance of Retail datasets that consist of Grocery Retail Data set (GRD), Retail Products Data set (RPD), and Customer Behavior Data set (CBD) are used to determine the effectiveness of the proposed model. The findings may indicate that the suggested approach is most precise during the procedure of being applied to detect and categorize the majority of the retail objects and, hence, may be optimally fit to be utilized in smart retail within the context of the real situation. The collected data of this study can be useful in other studies about the future trends in the field of retail automation aimed at boosting the level of positive customer experience and efficiency advantage.