Description: Participate in Mid-Autumn Festival activities
I. Introduction:
With the rapid development of the picture era, a large amount of text content is published and stored in the form of pictures in order to optimize layout and performance effects. This brings great convenience to the dissemination and security of the content and requires repetitive labor. .
OCR text scanning tools have gradually emerged, mainly to help users solve the problem of content editing.
2. What is OCR?
The full name of OCR is Optical Character Recognition, which means "optical character recognition technology".
1. The role of OCR:
The text in the image is recognized through OCR technology, extracted and converted into text format, and some structured text data is output.
2. The process of OCR realizing text recognition:
The principle of OCR is to use a scanner to convert the optical signal of the Chinese character document into an electrical signal through a charge-coupled device (CCD), and then convert it into a digital signal through an analog/digital converter and transmit it to the computer. The computer receives digital images of manuscripts and then recognizes the Chinese characters in these images.
3. OCR selection:
4. Application of OCR in daily life:
Today, with the rapid development of intelligent recognition technology, through careful review, I found that OCR recognition technology is applied in all aspects of life. The following are my actual application scenarios in life. See if you have the same experience?
(1). Children doing homework:
Usually, the elderly at home help their children with their homework. If they encounter something that they cannot do, they will use the "Homework Helper" software. Take photos of questions you don’t understand and upload them to the server. The server will recognize the text, search for the corresponding questions in the question bank, and return the query results to the interface.
(2). Self-media operation:
I often need to go to public accounts to post some copywriting and activities, and find better materials on the Internet. You should often see some paid copywriting materials downloaded from Baidu Wenku, etc. Sometimes, you can only rely on your own hands to type out the form. You can use WeChat’s built-in “text recognition” function for identification.
(3). Payment:
Before mobile phone QR code payment, people still paid with cash, but now people are paying with mobile phones in the streets and alleys. This new payment method has entered our lives and is gradually accepted by the public. Then we open Scan and scan the QR code of the merchant or individual to identify who it is? Even the more advanced facial recognition scan code payment, these are actually the most widely used methods of OCR application scenarios.
(4). Shared bicycles:
My work is relatively close to the company, so I scan shared bicycles and commute back and forth every day. Using WeChat to scan the QR code on shared bicycles is also an OCR application scenario.
The following is a summary of OCR application scenarios in life:
Next, combine the company's business and optimize the process to achieve the company's energy saving, cost reduction and efficiency improvement.
3. Company business:
The company is engaged in pet-related business. The company adheres to its mission of providing nutritious, healthy and safe food for pets and focuses on the research, development, production and sales of dog and cat food.
4. The company’s business pain points:
At present, OCR technology is being widely used. Tencent Cloud Text Recognition is a technology that uses OCR, which can help enterprises solve some business pain points and increase efficiency and reduce costs for the company.
1. Business scenario:
Scenario 1: The company's local promotion salesperson needs to go to the pet store to register and authorize the store to sell the company's products and expand the company's customers. The company's customer service department specifically needs to review the information submitted by the salesperson's local promotion. , at the same time, it also conducts assessment for business assessment.
Scenario 2: After purchasing, the supplier needs to submit the invoice to the company.
2. Business pain points:
- When a merchant is authenticated, the uploaded business license needs to be reviewed manually, which is very labor-intensive and a very boring and repetitive task.
- After the invoice is submitted, it needs to be reviewed manually by financial personnel, who often work overtime.
- By expanding manpower + overtime, the company's labor costs will increase.
Let’s reduce customer service financial operating costs through the practice of text recognition on Tencent Cloud.
5. Tencent Cloud text recognition practice:
Text recognition based on the deep learning technology of Tencent Youtu Lab intelligently recognizes the text content on the picture into editable text. OCR supports printed text recognition for cards and bills such as ID cards and business cards, as well as handwritten text recognition such as waybills. It supports the provision of customized services and can effectively replace manual information entry.
Text recognition based on the deep learning technology of Tencent Youtu Lab provides great advantages over other products when selecting products.
Based on Tencent's self-developed deep learning technology and massive data, it provides text recognition services for various scenarios and types such as cards, bills, printed and handwritten characters, and custom templates.
Tencent Cloud OCR is an excellent text recognition service with the characteristics of high accuracy, low error rate, fast recognition, etc., and can be applied to different scenarios. Using Tencent Cloud OCR can help us quickly process text information in images and improve work efficiency, and has been widely used.
1. Open related products:
It can be seen that Tencent Cloud has many OCR products, which can be widely used in many fields and become an auxiliary tool for efficient productivity.
In order to test the two scenarios mentioned in this article, the "Universal Text Recognition" type is selected.
Check "I have read and agree" to activate the text recognition product function.
Remember to use real-name registration for your account, otherwise you will be prompted that you need to use real-name registration. After passing the review, you can see that we have launched the "Text Recognition Service" for the first time. Each resource package has a number of free gifts, which allows us to conduct research and testing on the product, which is very considerate.
The first activation will give you 250 free opportunities, and there are 9 types of activation, which can be flexibly tested according to your own business needs.
2. Free test:
Tencent Cloud text recognition product family includes services such as general text recognition, general card recognition, bill and document recognition, text image enhancement, intelligent structured recognition, intelligent scanning and specific scene recognition. After activation, you can enjoy 1,000 times/month for free. Call amount.
3. Novice experience:
The official provides a variety of ways, according to which you can choose the appropriate method of use.
- Text recognition experience demo
- Visually call text recognition service
- One minute code access to server API
4. Online text recognition experience demo:
Simply click "Upload local file", select a business license, perform online analysis, and see the returned recognition results.
5. Visually call text recognition service - official debugging tool:
First test with the online debugging tool provided on the official website to see the effect. In "Signature String Generation", click "View Key" to view the ID and key.
Check the API ID and key. If the key is displayed, you need to verify it via SMS.
In "Online Call", fill in the input parameters with imageUrl and the signature string just generated, click "Initiate Call", you can see that the response result has returned data.
6. Build Node service analysis:
Tencent Cloud OCR has officially prepared a variety of ways to quickly integrate the Tencent Cloud Developer Tool Kit (SDK) corresponding to this interface into local projects. In order to demonstrate the function, this article uses Node's Koa framework for development.
| serial number | Bag | effect |
|---|---|---|
| 1 | too | Koa is an advanced framework for Node.js. It is based on the middleware mechanism of Node.js and provides a simpler and more flexible framework for building efficient and scalable web applications. |
| 2 | koa-bodyparser | Koa-bodyparser is a middleware of the web framework koa, which is used to parse the request body in HTTP requests, that is, parse the data in POST requests into ctx.request.body. |
| 3 | koa-router | koa-router is a middleware of koa, which also contains many middlewares. These middlewares are divided according to different routing paths through Layer objects. |
| 4 | tencentcloud-sdk-nodejs | The SDK used to access Tencent Cloud services can help you develop on Tencent Cloud using Node.js language. |
1. Initialize project:
mkdir orc-test
cd orc-test
npm init
# 一路按回车即可初始化一个package.json
# 安装插件
yarn add tencentcloud-sdk-nodejs@4.0.673 koa@^2.14.2 koa-bodyparser@^4.4.1 koa-router@^12.0.0
2. Write the code recognized by OCR:
const tencentcloud = require("tencentcloud-sdk-nodejs")
const OCRClient = tencentcloud.ocr.v20181119.Client
const Koa = require('koa');
const Router = require('koa-router');
const bodyParser = require('koa-bodyparser');
const client = new OCRClient({
credential: {
secretId: "AKIDyxpjjmxxxxxxxFdtx", # 使用自己的id
secretKey: "eFh0961yxxxxAQ", # 使用自己的密钥
},
// 产品地域
region: "ap-guangzhou",
})
// 实例化Koa对象 => app
const app = new Koa();
// 实例化路由对象 => router
const router = new Router();
app.use(bodyParser())
// 测试接口
router.get('/', async (ctx, next) => {
ctx.response.body = `<h1>Hello, Koa2</h1>`;
});
function getImg(ImageUrl) {
return client.BizLicenseOCR(
{
ImageUrl,
},
)
}
// 获取营业执照结果
router.post('/api/getBusiness', async (ctx, next) => {
const request = ctx.request.body
let result = await getImg(request.url);
ctx.response.type = 'application/json';
ctx.response.body = {
"code": '200', "message": '成功', "data": result };
});
app.use(router.routes()).use(router.allowedMethods());
// 在端口3000监听
app.listen(3000);
console.log('app started at port 3000...');
3. Build test url:
Just enter http://127.0.0.1:3000 of the get request in postman.
4. Test whether the business license can be parsed correctly:
In postman, enter http://127.0.0.1:3000/api/getBusiness in the post request. The post request parameter is url. You can see that the business license information can be returned.
5. Test whether the VAT invoice can be parsed correctly:
Replace the request image parsing function BizLicenseOCR with the function VatInvoiceOCR.
function getImg(ImageUrl) {
return client.VatInvoiceOCR(
{
ImageUrl,
},
)
}
The name of this method can refer to the following. Different types of pictures use different methods:
In postman, enter http://127.0.0.1:3000/api/getBusiness in the post request. The post request parameter is url. You can see the information that can return the VAT invoice.
6. Summary:
The following is my experience from 0 to 1, from entry to practice. It took me less than half an hour to complete the entire OCR experience process. I can feel that Tencent Cloud’s products are indeed simple and easy to use. At the same time, I also completed the requirements for business licenses, A survey of VAT invoices is completed.
7. Comparison before and after improvement measures:
6. Estimated import income:
- Using Tencent Cloud's OCR text recognition function can greatly simplify the business workflow.
- The previous purely manual operation has been changed to an automatic review mechanism. Those that cannot be identified or are identified incorrectly will be reviewed manually.
- It greatly facilitates the workload of customer service and financial personnel, and they no longer need to carry computers after get off work.
We made an estimated evaluation of the company's cost reduction strategy. The labor cost was reduced by about 30%, and the work efficiency was improved by more than 50%. It has become a basic public service, and there will be new business scenarios in the future. Can be online quickly.
Of course, research needs to be conducted based on the actual situation of the company. For example, if the cost of procurement is much greater than the cost of labor, it may need to be measured.
7. Summary:
The maturity of OCR technology has made content editing in the era of graphics and text easier. For business scenarios that often deal with text and pictures, text recognition and extraction tools based on OCR technology are indispensable tools for improving efficiency.
In the era of information society, a large amount of bills, forms, and document data are generated every day. If this data is to be transformed from manual processing to electronic information, it needs to be extracted and entered using OCR technology.