Shop Analytics-Customer behavior monitoring
Shop/cafeteria analytics system is suitable for broad range of applications some of them are human-object interaction, person counter, gender classification, age group classification and time spent in store by customer (Re-identification). Video surveillance system often rely on computer vision
2025-06-28 16:35:00 - Adil Khan
Shop Analytics-Customer behavior monitoring
Project Area of Specialization Artificial IntelligenceProject SummaryShop/cafeteria analytics system is suitable for broad range of applications some of them are human-object interaction, person counter, gender classification, age group classification and time spent in store by customer (Re-identification). Video surveillance system often rely on computer vision algorithms to automate surveillance tasks. By using the video streams, system will be able to do the image recognition. Through which we will perceive the required data. System will provide statistics of daily footfall. The cameras fitted at the entrance will detect in and out flow of persons. The gender of individuals will be identified and age will be marked through image recognition. The cameras used inside will provide the statistics of person occupancy in specific area and their interaction with objects at respective counter. We can deploy this system in any shop, mall and cafe for customer behaviour monitoring.
Project ObjectivesIndustry Objectives: The system can be deployed in any cafeteria, outlet and shop for monitoring customer behaviour, tracking and detection of customer.
Research Objectives:
Crowd Analysis: We can check the crowded area or rack the person interacted with, by which we can come to know the interest of customer.
Gender Classification: We can come to know the gender that visit the shop more likely.
Age group Classification: We can analyze the most visited age groups and their area of interest.
Person Detection and tracking: We can count and identify a person by tracking.
Academic Objectives: Our academic objective is to have grip on python, machine learning and natural language processing techniques.
Project Implementation MethodIn our project for customer behavior monitoring, we use Agile Methodology:
• In agile methodology, we will break our project into number of several spirits,
Our iteration will consist on same time period which is one spirit (2 weeks).
• At the end of each iteration we will deliver a working product.
• In our case, we will break our project into 20 release assuming each iteration of size 2 weeks.
• In 1st spirits, we basically focus on gathering data set, that is the core part for agile methodology. once the requirement gathering step is complete then we move further.
• In the next phase, we will focus to deliver the requirement features.
• Any required features that cannot be delivered in the first iteration will be taken up in the next iteration. • At the end of each iteration, we will deliver the working software that is incrementally enhanced and updated.
- In the first deliverable, we study the topics that are included in our Project:
- Person Detection and Tracking.
- Re identification of person
- Gender Classification.
- Age Group Classification.
- Human-object interaction
- In the second deliverable, we gather the data sets from the cafes.
- In the third deliverable, we develop the algorithms that are applied to our project.
- In the fourth deliverable, we test our project whether it works correctly or not.
- In the fifth deliverable, we deploy the project.
Mainly the direct customers of this project are outlets or cafes. They can get benefit directly from this project because they can find out how well they are performing by counting the number of customers enter in their outlets or cafes or in which rack/counter the number of customers will remain higher. So they can update their policy and strategies accordingly to increase their sales.
Aim of our project is to analyze customer behaviour that how many customers enter or leave the shop or cafe. We do this in two ways:
- person detection.
- Object detection and its interaction with human.
Input of our project is the video stream of camera. Expected outcome of our project is the counting of customers enter or exit the shop, the gender/age of the customer and the rack/counter where number of customers is higher
Technical Details of Final DeliverableOur project for customer behaviour monitoring is desktop application in which we used
- Python: The language used for overall implementation.
- Django: Frame work used for web development
- Yolov4 model: The module used for detection and tracking of persons.
- Fast RCNN: The algorithm used for human-object interaction.
- Caffemodel framework: Used for age and gender detection.
- MySQL database server: The server used to store results like age/gender and number of people, in database.
- Data Analytics Acceleration Library.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 20600 | |||
| camera | Equipment | 4 | 5000 | 20000 |
| printing | Miscellaneous | 5 | 120 | 600 |