TWI613601B - System and method for intelligent image recognition dynamic programming - Google Patents

Description

System and method for intelligent image recognition dynamic programming

The invention belongs to a dynamic route planning system and method using smart image recognition, and uses the photographic lens image to calculate the instantaneous load level of each area, and then sends it to the mobile device through the wireless network, suggesting that the customer can avoid the congestion according to the planned route. region.

More and more people in modern society like to shop in large stores or department stores, but it often happens that congestion occurs in a certain commodity area due to too many customers, which seriously affects the customer's shopping line and consumption experience; therefore, in order to solve the aforementioned headaches The problem is the creation of the "system and method of intelligent pedestrian image motion planning"; the purpose of these creations is to reduce the congestion of the venue, but the above creations provide information for the user to decide whether it is in the original The facility continues to wait in line or move to another facility, simply moving the customer from the congestion area to the open area, and does not plan the shopping route for the customer, and the overall efficiency of the space is not improved. The customer may have avoided a stagnation area but spent more time moving, which may waste more time in terms of the overall shopping route. The purpose of this creation is to plan the shopping route in real time with the customer's wish list, not only avoiding the smuggled goods area, but also changing the route immediately due to sudden crowds. It allows customers to have a smooth shopping route, convenience and time saving. In addition, the present invention relates to image monitoring in customer shopping analysis and shopping mall security. Control has a combination. Through the electronic image recognition technology, the customer's current behavior is analyzed, and after the intelligent product recommendation function is calculated, the customer's potential demand product is developed. Coupled with face recognition technology, once the image monitoring system recognizes suspicious people, it can immediately alert, effectively improving the safety management benefits of the mall.

In view of the shortcomings derived from the above-mentioned conventional methods, the inventors of the present invention have improved and innovated, and after years of painstaking research, they finally succeeded in developing and developing the system and method for dynamic planning of intelligent image recognition.

In order to achieve the above object, the present invention provides a system and method for intelligent image recognition dynamic planning, in which the customer traffic of the shopping mall grows year by year, and the phenomenon that the shopping route of the customer is often stagnant or not smooth due to the crowds, and the consumer experience is seriously damaged. It is inconvenient to manage the flow line with the market. Trying to give customers a smooth purchase route and developing customers' potential demand goods will also help the overall use of space and human flow line monitoring.

The invention comprises each camera, image capturing and lens control module, dynamic route planning module, customer historical consumption information module, customer purchase product behavior information module, smart product recommendation module, wireless network module, The suspicious person alarm module and the mobile device display module, when the customer enters the store, the customer shopping list is composed of the historical record module, the smart recommended product module and the goods manually selected by the customer. Through the camera to calculate the regional traffic load degree for different commodity areas and walkways, dynamically plan the shopping route that satisfies the customer list and send it to the customer mobile device via the wireless network. When there is an uncomfortable number of people in a specific commodity area, the shopping route is changed, and the customer mobile device that sends the message immediately enables the consumer to effectively avoid the blocked commodity area, and has a more comfortable space to purchase the product.

In addition, the smart recommendation product module can collect relevant behaviors and action information of the customer's selected products obtained through the camera, recommend the products that the customer may need, and develop the potential demand products of the customer. On the other hand, the suspicious person alarm module combined with the image monitoring system, combined with the database of the theft suspects, once the suspicious person is identified, the alarm is immediately raised, which greatly enhances the security management effect of the mall.

The invention can be applied to large shopping malls and department stores, etc., and effectively gives customers appropriate routes and product suggestions. When a specific commodity area is blocked, priority can be given to a smooth commodity area to enhance the customer's consumption experience in the store, and the shopping space. Effectiveness in use, store security management, and merchandise purchase rate can be effectively strengthened.

Therefore, the invention provides a system for intelligent image recognition dynamic planning, including a photographic lens, which is an image for real-time shopping route planning; a route planning and information collecting unit is responsible for data collection and information integration, and according to each of the captured images Information on the number of people in the commodity area, the load level of the customer, and the customer's shopping list, etc., to plan the best shopping route, including: image capture and lens control module, is an image capture device, and provides customers with the purchase of goods behavior information model. Group and dynamic route planning module for pedestrian image analysis and identification, customer consumption behavior identification and related information collection; customer historical consumption information module is to collect information about customers' past consumer goods information and consumption habits, and provide smart product recommendation module. For further analysis and reminding customers of possible daily necessities and matching promotions; the customer purchases the product behavior information module, which is a behavior collection and analysis system for customers to purchase goods, and implements relevant items according to customers' purchases. Behavior mode; smart recommendation product module, according to the customer The historical consumption information module and the customer purchase information collected by the product behavior information module, and through analysis, recommend the most suitable product to the customer, and in accordance with the product recommendation score interval purchased by the customer, recommend the product of the interval; Route planning module is based on goods The district acts as a node, and takes the walkway between the two commodity areas as a line, transforms the entire store into a dotted line graph, and then calculates it with a mathematical model to obtain a smooth shopping route; the mobile device display module is a dynamic route. The operation platform of the planning module; the wireless network module is the calculation result for transmitting the dynamic route planning module.

The relevant behavior mode is to stop the time, the adhesion of the product on the hand, pick up the number of related products, and the smart recommendation product module, including a calculation method to calculate the product recommendation score, which is calculated as: S = (c + t) * b * (1 + 0.1 * g) / average product score * 100%

Where S is the product recommendation score, c is the time for the customer to pick up the product, t is the number of times the customer picks up the product, b is the number of times the product is purchased by the customer, and g is the number of times the customer purchases a similar product.

The mathematical model of its dynamic route planning module is:

Where y ⁱ is the flow load fraction of each commodity area, x ⁱ is the score of the commodity in the commodity zone, the weight of each θ _i x ⁱ , h(x) is a function, W _i is the load weight of the commodity flow, and C ( θ ) is a cost function, in which the commodity score x ⁱ and the flow load fraction y ^{i of} each commodity area are supervised learning, and then the cost function C ( θ ) is used to train each commodity weight θ , and finally the function h is derived, and the flow of each commodity area is The load weight W _{i is obtained} by the customer taking the product score x ^{i of} this commodity area into h .

Deriving the weight of each commodity area W _i and the weight of the walkway L[a][b] , each node and line are weighted by W _i and L[a][b] , with the mall entrance as the starting node and the checkout counter as the final node. D is the score of the customer's shopping route, with the shortest path algorithm to minimize the customer's shopping route score D , in order to achieve customers avoiding the congestion area and smooth route shopping.

After the customer selects the required product and the recommended product, the module immediately calculates the weight of the node and the line through the Dijkstra's shortest path algorithm, and calculates the shortest path D from the entry (departure node) to the checkout counter (final node). The Dijkstra's shortest path algorithm is as follows: D _{t + 1} = D _t + W _b + L[a][b]

Where D is the customer's shopping route score, W _b is the commodity area flow load score, L[a][b] is the aisle flow load score of the commodity area a to the commodity area b, and if there is a commodity area or aisle suddenly appearing during the purchase of the customer When the weight is higher than the threshold, the module immediately updates the shopping route and sends a new route to the customer mobile device via the wireless network.

A method for intelligent image recognition dynamic planning Step 1. Take the photographic lens image for real-time shopping route planning; Step 2. Plan the best shopping route according to the information of the number of people in the commodity area, the flow load level and the customer shopping list; Step 3. If the load level of the commodity area is greater than the density threshold, dynamically correct the route and send the route to the mobile device of the party; Step 4. Collect the data related to the behavior and reaction of the customer when selecting the product; Step 5. Analyze statistics and recommend products that customers have potential needs.

In addition, in the security of shopping malls, in conjunction with the mall to steal the database of suspected criminals, once the suspicious individual is identified, the alarm will be immediately raised, the safety management effect of the shopping mall will be greatly improved, and the rate of theft of shopping mall items will be reduced.

The invention provides a system and method for intelligent image recognition dynamic planning, which has the following advantages when compared with other conventional technologies:

1. The invention can utilize image recognition technology for each commodity area and walkway Calculate the degree of flow load and make recommendations for shopping routes.

2. The invention can utilize image recognition technology to collect information and analyze the behavior of customers when purchasing goods, and provide customers with potential demand products.

3. The present invention can dynamically change the shopping route reference according to the flow of people.

4. The present invention includes a mobile device interface, and the customer can immediately receive updated shopping routes and instant offer information.

5. The invention can use the face recognition technology to detect whether there are suspicious persons in the mall and has an immediate alarm function.

110‧‧‧Photographic lens

120‧‧‧ Route Planning and Information Collection Unit

121‧‧‧Image capture and lens control module

122‧‧‧Customer History Consumer Information Module

123‧‧‧Customer purchase product behavior information module

124‧‧‧Smart Product Recommendation Module

125‧‧‧Dynamic Route Planning Module

130‧‧‧Mobile device display module

140‧‧‧Wireless network module

150‧‧‧ Suspicious Person Alarm System

S210~S250‧‧‧ Process

The detailed description of the present invention and the accompanying drawings will be further understood, and the technical contents of the present invention and the functions thereof can be further understood. FIG. 1 is a structural diagram of a system and method for intelligent image recognition dynamic planning according to the present invention. FIG. 2 is a flow chart of a system and method for intelligent image recognition dynamic planning according to the present invention.

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The present invention is further described with reference to the accompanying drawings. Please refer to FIG. 1 , which is a structural diagram of a system and method for intelligent image recognition dynamic planning according to the present invention, including a photographic lens 110 for capturing images for real-time shopping route planning; The Route Planning and Information Collection Unit 120 is responsible for data collection and information integration, and relies on the number of people in each product area in the image. The information on the level of the load and the customer's shopping list, and the best shopping route, including the image capture and lens control module 121, is an image capture device, and provides a customer purchase product behavior information module 123 and The dynamic route planning module 125 performs human flow image analysis and identification, customer consumption behavior identification and related information collection; the customer historical consumption information module 122 collects information about customers' past consumer product information and consumption habits, and provides a smart product recommendation module. 124 for further analysis and reminding customers of the daily use of goods and matching promotions that may be missed; the customer purchase product behavior information module 123 is a behavior collection and analysis system for customers to purchase goods, and according to the items purchased by customers, The relevant behavioral mode is implemented; the smart recommendation product module 124 is based on the information collected by the customer history consumption information module 122 and the customer purchase product behavior information module 123, and is recommended to the customer for the most suitable product through analysis, and in accordance with The product is recommended to purchase the product recommendation score interval, and the product recommended in this interval; The state route planning module 125 takes the commodity area as a node, and takes the walkway between the two commodity areas as a line, converts the entire store into a dotted line graph, and then calculates with a mathematical model to obtain a smooth shopping route; The mobile device display module 130 is an operation platform for the dynamic route planning module; the wireless network module 140 is a calculation result for transmitting the dynamic route planning module 125; in addition, in the security of the shopping mall, there is a suspicious person alarm The system 150, in conjunction with the mall stealing the database of the suspects of the former, once the suspicious person is identified as an immediate alarm, greatly enhances the security management effect of the mall and reduces the rate of theft of the goods in the mall. The relevant behavioral mode is to stop the time, the degree of adhesion of the product on the hand, and the number of related items.

The route planning and information collection unit 120 is a data transmission and reception and computing center, including a computing core, a controller and a memory, and the like, and a peripheral module such as a mobile device display module 130, a wireless network module 140, and a suspicious person alarm module 150. Group, let this unit use the acquired image and related data as an internal program Computing and overall system control center. The image capturing and lens control module 121 is an image capturing device of the dynamic route planning module 125 of the customer shopping, and the captured image is mainly used by the customer to purchase the product behavior information module 123 and the dynamic route planning module 125 for image analysis of the human flow. Identification, customer consumption behavior identification and related information collection. The mobile device display module 130 is a customer operation platform of the customer's shopping dynamic route planning module 125, and is provided for the customer to input the desired product, the customer purchases the product recommended by the smart product recommendation module 124, and receives the product through the wireless network module 140. The planned route of the dynamic route planning module 125. The wireless network module 140 is an existing and future telecommunication wireless communication network interface. The telecom wireless communication network transmits the dynamic route planning module 125 to calculate the result, so that the other end of the network receives the mobile device. The planned route, in addition to the customer's potential consumer goods through the mobile web interface.

The customer history consumption information module 122 collects information about the past consumer product information and consumption habits, and provides the smart product recommendation module 124 for further analysis and reminds the customer of the daily use of the product and the matching preferential activities. The customer purchase product behavior information module 123 collects and analyzes the behavior of the customer when purchasing the product. When customers purchase items, they will have relevant behavior patterns, such as stopping time, sticking the goods on the hands, picking up the number of related items, etc., plus basic customer information such as age, gender, income, etc. The above related information can be collected by the image capturing and lens control module 121 for analysis of related modules. The information collected by the smart recommendation product module 124 for the integrated customer history consumption information module 122 and the customer purchase product behavior information module 123 is analyzed and recommended to the customer for the most suitable product. The smart recommendation product module includes a calculation method to calculate the product recommendation score, which is calculated as: S = (c + t) * b * (1 + 0.1 * g) / commodity average score * 100%

Where S is the product recommendation score, c is the time for the customer to pick up the product, t is the number of times the customer picks up the product, b is the number of times the product is purchased by the customer, and g is the number of times the customer purchases a similar product.

The dynamic route planning module 124 mainly uses the commodity area itself as a node, and the walkway between the two commodity areas serves as a line, and converts the entire store into a dotted line chart. Firstly, a mathematical model is established. The previous product score x _i and the flow load score y _{i of} each commodity area are used for supervised learning. Then, the cost function C (θ) is used to train each commodity weight θ , and finally the function h is derived. The traffic load weight W _i of each commodity area is obtained by the customer taking the product score X _i of the commodity area into h , and the aisle weight L[a][b] is the total number of carts in the aisle and the customer adhesion in the aisle. . The mathematical model of its dynamic route planning module is:

Wherein y ⁱ is the flow load fraction of each product region, x ⁱ is the fraction of commodities trade area, θ _i for each x ⁱ of the weight, h (x) is a function, W _i is the flow load weight goods area weight, C (θ) is a cost function, in which the commodity score x ⁱ and the flow load fraction y ^{i of} each commodity area are supervised learning, and then the cost function C ( θ ) is used to train each commodity weight θ , and finally the function h is derived, and the flow of each commodity area is The load weight W _{i is obtained} by the customer taking the product score x ^{i of} this commodity area into h .

Deriving the weight of each commodity area W _i and the weight of the walkway L[a][b] , each node and line are weighted by W _i and L[a][b] , with the mall entrance as the starting node and the checkout counter as the final node. D is the score of the customer's shopping route, with the shortest path algorithm to minimize the customer's shopping route score D , in order to achieve customers avoiding the congestion area and smooth route shopping.

After the customer selects the required product and the recommended product, the module immediately calculates the weight of the node and the line through the Dijkstra's shortest path algorithm, and calculates the shortest path D from the entry (departure node) to the checkout counter (final node). The Dijkstra's shortest path algorithm is as follows: D _{t + 1} = D _t + W _b + L[a][b]

Where D is the customer's shopping route score, W _b is the commodity area flow load score, L[a][b] is the aisle flow load score of the commodity area a to the commodity area b, and if there is a commodity area or aisle suddenly appearing during the purchase of the customer When the weight is higher than the threshold, the module immediately updates the shopping route and sends a new route to the customer mobile device via the wireless network.

Please refer to FIG. 2 , which is a flowchart of a system and method for intelligent image recognition dynamic planning according to the present invention. The flow is as follows: Step 1. S210 takes a photographic lens to take an instant shopping route planning; Step 2. S220 according to each Plan the best shopping route by information such as the number of people in the commodity area and the load level of the customer and the customer's shopping list; Step 3. If the load level of the commodity area is greater than the density threshold, the route is dynamically corrected and the route is sent to the mobile device of the party. Step 4. S240 collects the data related to the behavior and reaction of the customer when selecting the product; Step 5. After analyzing and counting, S250 recommends the goods of the customer's potential demand.

In addition, in the security of shopping malls, in conjunction with the mall to steal the database of suspected criminals, once the suspicious individual is identified, the alarm will be immediately raised, the safety management effect of the shopping mall will be greatly improved, and the rate of theft of shopping mall items will be reduced.

The detailed description of the present invention is intended to be illustrative of a preferred embodiment of the invention, and is not intended to limit the scope of the invention. this In the scope of the patent.

To sum up, this case is not only innovative in terms of technical thinking, but also has many of the above-mentioned functions that are not in the traditional methods of the past. It has fully complied with the statutory invention patent requirements of novelty and progressiveness, and applied for it according to law. Approved this invention patent application, in order to invent invention, to the sense of virtue.

110‧‧‧Photographic lens

120‧‧‧ Route Planning and Information Collection Unit

121‧‧‧Image capture and lens control module

122‧‧‧Customer History Consumer Information Module

123‧‧‧Customer purchase product behavior information module

124‧‧‧Smart Product Recommendation Module

125‧‧‧Dynamic Route Planning Module

130‧‧‧Mobile device display module

140‧‧‧Wireless network module

150‧‧‧ Suspicious Person Alarm System

Claims (5)

A system for intelligent image recognition dynamic planning, comprising: a photographic lens that captures images for real-time shopping route planning; a route planning and information collecting unit that is responsible for data collection and information integration, and according to which the products in the image are captured Information on the number of people in the district, the level of traffic load, and the customer's shopping list, to plan the best shopping route, including: image capture and lens control module, which is an image capture device, and provides customer purchase information behavior module And the dynamic route planning module for the analysis of the human flow image analysis, the identification of customer consumption behavior and the collection of related information; the customer historical consumption information module collects relevant information on the consumer information and consumption habits of the customers, and provides the smart product recommendation module. Further analysis and remind customers of the daily use of goods and collocation activities that may be missed; the customer purchases the product behavior information module, which is the behavior collection and analysis system for customers when purchasing goods, and performs relevant behavior according to the items purchased by customers. Mode; smart recommendation product module, according to the customer calendar The consumer information module and the customer purchase information collected by the product behavior information module, and through analysis, recommend the product that is most suitable for the customer, and further recommend the product in the range according to the product recommendation score interval purchased by the customer; The route planning module uses the commodity area as a node, and takes the walkway between the two commodity areas as a line, converts the entire store into a dotted line graph, and then calculates with a mathematical model to obtain a smooth shopping route; The display module is an operation platform of the dynamic route planning module; The wireless network module is a calculation result of transmitting the dynamic route planning module. For example, the system for intelligent image recognition dynamic planning described in claim 1 of the patent scope, wherein the related behavior mode is the time of stopping, the adhesion of the product on the hand, and the number of times the related product is picked up. For example, the system for intelligent image recognition dynamic planning described in claim 1 wherein the smart recommendation product module includes a calculation method for calculating a product recommendation score, which is calculated as S = (c + t) * b * (1 + 0.1 * g) / average product score * 100%, where S is the product recommendation score, c is the time for the customer to pick up the product, t is the number of times the customer picks up the product, b is the number of times the customer purchases the product, g is The number of times a customer purchases similar items. For example, the system for intelligent image recognition dynamic planning described in claim 1 of the patent scope, wherein the mathematical model is: Where y ⁱ is the flow load fraction of each commodity area, x ⁱ is the score of the commodity in the commodity zone, the weight of each θ _i x ⁱ , h(x) is a function, W _i is the load weight of the commodity flow, and C ( θ ) is a cost function, in which the commodity score x ⁱ and the flow load fraction y ^{i of} each commodity area are supervised learning, and then the cost function C ( θ ) is used to train each commodity weight θ , and finally the function h is derived, and the flow of each commodity area is The load weight W _{i is obtained} by the customer taking the product score x ^{i of} this commodity area into h . A smart image recognition dynamic planning method includes: Step 1. Taking a photographic lens to take an instant shopping route planning; Step 2. According to the number of people in each commodity area and the load level of the person and the customer shopping Single information, plan the best shopping route; Step 3. If the load level of the commodity area is greater than the density threshold, dynamically correct the route and send the route to the mobile device of the party; Step 4. Collect the behavior of the customer when selecting the product Steps related to actions and reactions Step 5. After analyzing and statistic, recommend the products of potential customers' needs.