Smart Retail Solution

The solution to collect and analyze customer behaviour data

Ninh Trong Hoang
Nguyen Tuan Dat
Le Duc Nguye
Nguyen Thi Huong Thao

The Smart Retail Solution project comprises a variety of features and functionality, including queue management, employee management, heatmaps, indoor maps, and people counting. The Smart Retail Solution Project, on its whole, is a web-based application that provides data statistics from numerous smart cameras acquired with the aid of AI, Smart Camera, and Machine Learning Technologies. The TMA teams had already worked on a few features, along with some unfinished features that we are required to carry out. 

In detail, our team remains as two sub-teams that work in different fields of the industry, but still serve the same scope of the project:

The AI team: Responsible for developing algorithms, models for machine learning, image processing, feature extraction and analysis. 

At first, we were provided with some AI Basic training sessions, which included: machine learning, computer vision, and deep learning. While the training session is executed, the scope of the AI team has been updated from AI Training for Behavior and heatmap generating to AI Training for Parking Lot. 

For better understanding, the AI Training for Parking Lot task is to develop the new AI model, as well as upgrade and maintain the feature from the existing one that had been designed previously by the industry. The model can then be used for multiple usages: detecting vehicles, parking slot recognizing, LCD display for parking guide, etc., with the technology provided to one or multiple camera sensors on the parking area.

From the perspective of the business strategist, the parking detection solution is always considered to be one of the most important strategies that can affect the retail industry. The benefits of a great parking system play a huge role in bringing profit toward the retail market, such as creating a positive customer experience, returning customers, and greater revenues. Therefore, even though the task of our team has been changed, it is still one of many major tasks related to the scope of the project 

The Web team: Responsible for planning, designing and building a full-stack feature for the current website. 

By applying the knowledge of the training sessions that the team has completed in the Capstone A, such as: business analysis, conducting market research, designing UX/UI for simple applications, two members of the team (Thao & Hoang) are required to plan, design and implement a feature called Queue Analysis. 

Queue Analysis is an extended feature that belongs to the Smart Camera System website. The feature will be associated with other features such as Demographic, Customer Behavior, and Heat-map Analysis. All of the mentioned features belong to a more general category, Customer Analysis. 

Queue Analysis is expected to be able to show the numbers of customers at different queues at the store. Along with the number of customers, each queue will have its own status corresponding to the number such as Empty, Normal or Overloaded. In addition, as planned, the feature will also allow users to select real-time cameras in order to observe the insights that relate to the queues that are being captured within the cameras’ frames. A pop-up notification will be used to alert users about the status of a queue if it is overloaded.