Camera-Based Checkout System Using Facial Recognition and Motion Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing 'Just Walk Out' checkout mode using RFID technology is labor-intensive and costly due to the need for adhesive RFID tags on each item, which can be damaged or cause signal shielding issues, especially with metal items, leading to potential losses for merchants.
Innovation Solution
A method and system that uses facial recognition and hand motion tracking with convolutional neural networks to identify customers and associate their actions with items in real-time, generating shopping lists without the need for RFID tags, employing cameras and machine learning models to track customer movements and classify items accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If RFID tags are adhered to each item, then item identification is enabled, but the workload for workers increases and cost increases
Solution Approach 1:
The patent extracts the identification function from the item itself (by removing RFID tags) and relocates it to the camera system. The camera captures images of items, and the system identifies items through image recognition and matching with a database, eliminating the need for physical RFID tags on each item.
Solution Approach 2:
The patent replaces the mechanical RFID tagging system with an optical camera-based identification system. Instead of using radio frequency signals and physical tags, the system uses cameras to capture images and computer vision algorithms to identify items, thereby reducing manual workload.
2Reliability
If RFID tags are adhered to each item, then item identification is enabled, but the cost increases
Solution Approach 1:
The patent removes the costly RFID tags from each item and replaces them with a centralized image database and camera system. The identification cost is shifted from per-item tags to a one-time database construction and camera system deployment.
Solution Approach 2:
The patent replaces expensive, durable RFID tags with a more economical camera-based system. While cameras are not disposable, the overall system cost is reduced by eliminating the need to purchase and maintain RFID tags for every item in the inventory.
3Extent of automation
If RFID tags are used, then checkout is automated, but signal shielding problems occur with metal items
Solution Approach 1:
The patent replaces the electromagnetic RFID signal system with an optical camera-based identification system. Cameras capture images of items, and computer vision algorithms identify items regardless of whether they are metallic or non-metallic, eliminating signal shielding issues.
Solution Approach 2:
The patent changes the identification parameter from radio frequency signals (which are blocked by metal) to optical images (which are not affected by material composition). This parameter change allows the system to identify both metallic and non-metallic items equally effectively.
4Reliability
If RFID tags are adhered to items, then item tracking is enabled, but tags may fall or be damaged causing loss
Solution Approach 1:
The patent extracts the identification function from physical tags that can fall or be damaged and relocates it to a centralized database with item images. The camera system captures images and matches them against the database, eliminating the need for physical tags on each item.
Solution Approach 2:
The patent creates a digital copy (image database) of each item and uses this copy for identification purposes. Instead of relying on physical tags that can be damaged, the system uses durable digital images stored in a database, which can be repeatedly accessed without degradation.
Data Source
AI summary
The present invention discloses a method, a device and a system for checkout, and belongs to the technical field of computer. The checkout method comprises: identifying a pre-registered customer to acquire an identity information of the customer, the identity information containing face data and a payment account of the customer; tracking the customer whose identity information has been acquired, in a shopping place in real time, and acquiring a position of the customer; judging whether the position of the customer is consistent with a position of an item in the shopping place, and if so, associating the customer with a take-up action or a put-back action aiming at the item, and generating a shopping list of the customer after identifying the take-up action or the put-back action, and the item at which the take-up action or the put-back action aims; and performing checkout of the shopping list. The present invention further discloses a checkout device comprising a top camera, a forward camera, a downward camera and a server. The present invention further discloses a checkout system comprising a client and a checkout device. According to the above-mentioned technical solutions of the present invention, the situations that workload is heavy due to the adhesion of Radio Frequency identification (RFID) tags and RFID tags are easily be damaged may be avoided.