PRICE IMPROVEMENT USING FACIAL RECOGNITION SYSTEM

Abstract

This study aims to explore how pricing can be optimized using artificial intelligence techniques, including machine learning and deep learning.

Traditional methods of setting price for goods and services are some of the main methods that organizations have relied on for long time. The most notably method is cost-based pricing, where the price is calculated by adding direct and indirect costs and adding a specific profit margin that ensures sufficient returns for the organization. Competition-based pricing is also used, where the price is determined based on market and competitor prices without placing significant emphasis on actual costs. This is appropriate in markets characterized by product convergence and multiple competitors. Another traditional method is perceived value-based pricing, which relies on the customer's assessment of the value of a product or service. Psychological pricing is used, where product pricing is based on methods that influence consumer perceptions. Traditional methods also include promotional pricing, which is used for temporary offers to stimulate demand, especially during recessions or when launching new products. Geographic pricing, which takes into account differences in costs, taxes, and market conditions across different geographic regions, allowing for flexibility in distribution and marketing. Although these methods were effective in previous periods, technological and behavioral developments is driving many organizations to adopt more sophisticated pricing methods that respond to changing market conditions.

Facial expression recognition plays an important role in several fields, among them optimizing the prices of goods and services. Facial recognition systems generally consist of three main stages: preprocessing, feature extraction, and classification.

Preprocessing is a vital step in improving face recognition performance. It involves enhancing image quality through operations like clarity adjustment, scaling, and noise removal. This step also eliminates irrelevant details (e.g., ears) and prepares the image for accurate recognition by applying techniques such as alignment, normalization, binarization, and standardization.

Feature extraction focuses on extracting key facial features — like the eyes, nose, and mouth—and their geometric arrangement to classify expressions. Each face has a distinct structure that enables recognition. Techniques such as eigenfaces and scale-invariant feature transform are used for accurate feature extraction. Facial emotions are conveyed through the activation of specific muscle groups, revealing complex information about a person’s mental state. Machine learning and deep learning techniques are used to recognize and classify these expressions by training models on labeled facial images.

The goal of price optimization is to find the best pricing strategy that leads to setting the appropriate price, maximizing profit, and meeting customer needs. This can be achieved by relying on customer behavior through facial expressions, as the face is a key feature in expressing emotions. By analyzing customer facial expressions relative to the prices of goods and services offered in retail stores (supermarkets), the store owners understand better the customer reactions to prices.

The results of analyzing customer facial expressions, both positive and negative (such as happiness, sadness, anger, surprise, fear, and disgust), provide store owners with accurate insights into customers' emotional feelings as they interact with products or services. These analysis enable store owners to precisely meet customer needs, design personalized offers and services, and enhance the shopping experience. This, in turn, leads to increased sales, builds trust between the store and customers, enhances customer loyalty and satisfaction, and increases profits. It also enables store owners to make accurate decisions based on available data, creating a competitive advantage in a market characterized by constantly changing customer tastes.