How to Use Alibaba Cloud to Build a Perception Retail Platform

Abstract: 1. Background Introduction The so-called perceptual retailing refers to judging the level of interest of customers in products according to the emotional changes of store customers and the position of customers on the shelves of products in the store, and adjusting marketing strategies online in real time. Help enterprises to grasp customer feelings at the first time, issue marketing strategies at the first time, and get feedback on marketing effects at the first time.

Original address: http://click.aliyun.com/m/43688/

1. Background Introduction The

so-called perceptual retail refers to judging that customers are interested in products based on the emotional changes of customers in the store and the position of customers in the store where they stay on the product shelves degree, adjust marketing strategies online in real time. Help enterprises to grasp customer feelings at the first time, issue marketing strategies at the first time, and get feedback on marketing effects at the first time. To complete such a system construction, traditional enterprises and retailers often need to invest a lot of manpower, material resources and time, and some traditional enterprises often do not have such manpower and material resources, but what should be done urgently? Today, I will use real cases to teach you how to use Alibaba Cloud to build a perceptual retail platform. This real case comes from Alibaba Cloud helping three squirrels to complete the construction of the perception platform. The rendering is as follows, and the sensitive data has been processed.
Picture description

2. Overall architecture

Alibaba Cloud products are used in this construction. The Alibaba Cloud products and services involved are: iDST image recognition, DataV data visualization, object storage OSS, Alibaba Cloud live broadcast, ECS, RDS, SLB and other products
Architecture Diagram As follows:
Picture description Introduction to

business structure:
1. First, push the real-time monitoring data of the store camera to the Alibaba Cloud live broadcast platform by way of streaming;
2. Use the screenshot function of the Alibaba Cloud live broadcast platform to take screenshots of the monitoring video and store the pictures in the Alibaba Cloud live broadcast platform. The object is stored on OSS;
3. The development back-end application calls the iDST image recognition interface, sends the image to iDST for analysis and processing, and obtains the returned data;
4. Process and process the image data returned by iDST, and then send the data to the front-end processing application;
5. The front-end application follows up with the customer-related location coordinate data returned by the back-end application to draw a heat map, and the front-end needs to complete the control 6. Use the RTMP live broadcast component in the
DataV data visualization platform to display the real-time monitoring of the store, and use the data report component in the DataV data visualization platform to display the real-time transaction data of the store, etc.;
7. The front-end console issues real-time promotions Strategy, by calling the electronic price tag API interface, modify the price of store goods in real time.

3. Implementation steps

1. Build a live broadcast platform

1.1 Log in to the Alibaba Cloud console, enable the Alibaba Cloud live video function, and then enter the Alibaba Cloud live video function;
picture description

1.2 Create a new domain name After

entering the live video console, click "Domain Name Management" -> New Domain name, add live domain name (note that this domain name needs to be filed in advance, otherwise it will not pass); picture
description After the successful creation of a domain name, a domain name record will appear, and then click "Manage"; picture description In order to protect our content resources from being illegal If the site is downloaded and stolen, URL authentication needs to be configured. After entering the management page, you can see the relevant information of your live streaming, and then we click "Authentication Configuration"; picture description










After entering the authentication configuration page, you need to configure the streaming address, and the URL format can be filled in according to the suggested format. Continue to fill in the authentication KEY, and you can customize a character or number. It can be understood that this KEY is a string used for encryption. The validity period refers to the effective period of the live broadcast, after which the URL will automatically become invalid. After filling in, click Generate to get the URL for authentication. This URL is also our push stream address, copy and save it. For example, if you have 6 cameras, you need to configure 6 different authentication URLs;
picture description

Picture description

1.3 Enable the screenshot function Click "Screenshot Management"
in the live video console to enable the screenshot function. Note that the premise of enabling the screenshot function is to enable the object first. Store OSS, if OSS has been opened, it is not required. Because the screenshots of all the videos here are stored in the OSS;
picture description The

above -mentioned live broadcast platform has been built, and the purpose is to push the monitoring video data of the camera to the live video platform of Alibaba Cloud in real time, using the live video platform The screenshot function saves the picture in OSS. By default, a picture will be captured every 5 seconds, and the name of the last picture will be overwritten, so the content of the picture will be different each time you access it. Of course, if you want to save each screenshot, you can also choose to save it as a different file name each time. .

2. Configure camera streaming

2.1 Purchase several webcast cameras that support streaming function, the unit price on the market is about 200 yuan;

2.2 Power on the camera and connect to the network, each camera is equipped with a fixed IP by default, and the IP segment is 192.168.1.x network segment, you can log in to the management console by accessing the address of the camera with a web browser;

2.3 Configure push streaming After
entering the camera management console, click Remote Settings - Network Settings - RTMP Streaming; proceed as shown in the following figure Configure, disable unnecessary options, fill in the authentication URL address just obtained in the main stream configuration, then click save and refresh, when you see that the status is connected, it means that the push stream configuration is successful;
picture description

2.4
Confirming the camera has been configured to push the stream. Now we need to log in to the Alibaba Cloud live broadcast platform to check whether the push stream is successful. Enter the live video console, click Stream Management, and find that there is an additional video live streaming record, indicating that the streaming is successful. We can click the live broadcast address to obtain the playback address. Copy the playback address and put it on the RTMP of the DataV data visualization platform. Used in the player component;
picture description The

picture description is

OK. So far, we have transmitted the monitoring data to the Alibaba Cloud live video platform by means of streaming, and converted the monitoring video screenshots into picture data through the screenshot function of the live video platform. , which is stored in the object storage OSS.

3. Back-end application development The

development process is omitted. Here we will talk about the ideas. As mentioned earlier, we have converted the surveillance video data into image data. Next, we need to use the image recognition interface of Alibaba iDST to complete our image recognition. processing needs. The development process is omitted here, because each enterprise has different development needs. For example, our scenario is to analyze how many customers are in this picture, what are their ages, how many customers have smiling faces, real-time location information of customers in the store, and so on. Please refer to the detailed interface document provided by iDST for the specific functions. Our back-end application architecture uses two ECS servers plus an internal load balancing SLB and a high-availability RDS database.
The official image recognition interface address of iDST is provided here:
http://dev-ai.yunos.com/htmlpages/mais#/document/facedetect
Image description

4. Front-end development

The development process is also omitted. Here I will give you some ideas. When we complete the development of the back-end application and get the data we want, we also need the cooperation of the front-end to display the data. For example, we can display the data of customers in the store. The location is displayed by heat map, and the location of customers with smiling faces can also be displayed by heat map, so that the heat map distribution of customers with different emotions can be obtained. Of course, other requirements can also be flexibly implemented. In addition, we also need to develop some promotional console interfaces in the front end, mainly for discounts and promotions for products in specific regions. The product name and price of the promotion are sent to the back-end application, and then the back-end application calls the API interface of the electronic shelf sign, and the electronic shelf sign system sends the latest product price information to the electronic shelf sign AP device of each store. , and finally the electronic shelf sign AP sends the price to each commodity shelf sign device. Finally, it is possible to modify the commodity price of the store in real time. Our front-end applications are also deployed on two Alibaba Cloud ECS servers, plus a set of load balancing SLBs to achieve a high-availability cluster architecture.

5. Create a big screen of real-time marketing data

. Now that the front-end and back-end data are ready, we are about to start building our big screen of marketing data. Here we use the DataV data visualization platform of Alibaba Cloud's big data platform. It provides a very rich variety of data report components and controls.

5.1 Activate the DataV data visualization function
Enter the Cloud console, click Big Data (Data Plus) - DataV Data Visualization, and purchase the activation service;
picture description

5.2 Create a large data screen
Enter the DataV data visualization console and click "New Visualization" , and quickly complete the development of the large data screen by selecting the template on the left;
picture description

Picture description

5.3 Configuring the content of the large screen
After the big screen is created, we need to fill in the content of our data big screen according to business scenarios and various data. I will not introduce how each component is used in detail here. I will only introduce a few of the controls I use here. Among them, my real-time monitoring video is realized through the RTMP video stream player in "Auxiliary Graphics".
Image description The iframe component is

also used to display our front-end content through this component, because our front-end page implements functions such as promotion console and heat map.
Image description

Use the "Vertical Basic Bar Chart" in the "General Chart" to display the real-time sales and inventory of the three squirrel feeding stores.
Image description

Use the "arc bar chart" in the "regular chart" to show the age distribution of customers.
Picture description

Use the "Target Proportion Pie Chart" in the "General Chart" to display the sales target ratio of the store.
Picture description

Of course, there are many, many other fun controls, just drag and drop a control and configure it, it's OK, so easy. If you are interested, you can try it out and create a big data screen of your own.

At this point, we have completed the construction of the perception platform, is it very simple?
Picture description

4. Personal perception

At present, many small and medium-sized enterprises do not have professional artificial intelligence R&D teams and big data teams. When they want to create a high-quality product, they often have to stay away. Now Alibaba Cloud has opened these core technologies to the whole society through interfaces and cloud services. It makes the impossible possible before, and makes high-end technology more inclusive. Enterprise developers do not need to master artificial intelligence technology and image recognition technology to make high-end black technology products. Small and medium-sized enterprises can invest in lower hardware costs, fewer developers, and make the most competitive products in the shortest time. In the future, how to make good use of image recognition technology, language recognition technology, artificial intelligence, big data and other technologies will become the key elements of the core competitiveness of enterprises. Alibaba Cloud frees enterprises from the cumbersome construction and maintenance of physical computer rooms, provides agile, flexible and reliable cloud computing services, and provides the most advanced artificial intelligence and algorithm technology, big data, image recognition, speech recognition and other technologies to the whole society. It is also the original intention of Alibaba Cloud to empower enterprises to allow enterprises to focus on their areas of expertise based on reliable operational capabilities and practical experience.