Dynamsoft Capture Vision for Node

This is a Node.js wrapper for Dynamsoft Capture Vision, enabling you to read barcodes, recognize partial label text, capture documents, and perform other advanced vision tasks.

Install the SDK

npm i dynamsoft-capture-vision-for-node@3.2.5002 -E
# In many tasks, we used AI models.
# For simplicity, we will also install this package.
npm i dynamsoft-capture-vision-for-node-model@3.2.5001 -E

Getting Started

Take barcode reading from an image as an example.

const { LicenseManager, CaptureVisionRouter, EnumPresetTemplate } = require('dynamsoft-capture-vision-for-node');

// You can get your reading barcodes trial license from
// https://www.dynamsoft.com/customer/license/trialLicense?product=dbr&package=desktop
// You can get your recognizing partial label text, capturing documents or other feature trial license from
// https://www.dynamsoft.com/customer/license/trialLicense?product=dcv&package=desktop
// The current license is valid for only one day.
LicenseManager.initLicense('DLS2eyJvcmdhbml6YXRpb25JRCI6IjIwMDAwMSJ9');

(async()=>{
  // you can get the image from https://github.com/Dynamsoft/capture-vision-nodejs-samples/blob/main/AllSupportedBarcodeTypes.png
  // The second parameter `templateName` tells the SDK how to process this image.
  let result = await CaptureVisionRouter.captureAsync('./AllSupportedBarcodeTypes.png', EnumPresetTemplate.PT_READ_BARCODES_READ_RATE_FIRST);
  // refer to https://www.dynamsoft.com/capture-vision/docs/server/programming/cplusplus/api-reference/capture-vision-router/auxiliary-classes/captured-result.html?product=dbr&lang=cplusplus
  // or run `console.log(require('node:util').inspect(result, false, null))` to see details
  for(let item of result.barcodeResultItems){
    console.log(item.text);
  }

  // Terminate workers so you can exit the process.
  // Don't call it if you still need to use the SDK.
  await CaptureVisionRouter.terminateIdleWorkers();
})();

If your trial license has expired, please visit private trial for reading barcodes or private trial for recognizing partial label text, capturing documents or other feature.

Note

Guides for tasks such as Label Recognition and Document Capturing are under development. You can refer to the C++ document while writing code for Node.js. Or contact us for assistance.

Template Customization

The functionality of DCV largely depends on the choice of template. Dynamsoft offers preset templates in EnumPresetTemplate.XXXX. Here is the part about barcodes:

/** compatible with "read-barcodes" */
PT_READ_BARCODES = "ReadBarcodes_Default",
/** Represents a barcode reading mode where speed is prioritized. */
PT_READ_BARCODES_SPEED_FIRST = "ReadBarcodes_SpeedFirst",
/** Represents a barcode reading mode where barcode read rate is prioritized. */
PT_READ_BARCODES_READ_RATE_FIRST = "ReadBarcodes_ReadRateFirst",
/** Represents a barcode reading mode for single barcode code detection. */
PT_READ_SINGLE_BARCODE = "ReadSingleBarcode",

How to Specify the Barcode Formats

Below is an example illustrating how to modify the target barcode formats for PT_READ_BARCODES_READ_RATE_FIRST.

1. Copy the SDK's Templates\DBR-PresetTemplates.json to your directory. This file contains all the preset templates related to barcode reading.

2. Search BarcodeFormatIds and choose the one you need.

   e.g., If you only want to read specific barcode formats and ensure a high recognition rate in PT_READ_BARCODES_READ_RATE_FIRST, you can modify the BarcodeFormatIds object in the task-read-barcodes-read-rate task.

3. Suppose you only want to recognize QRCode and DataMatrix; you can change BarcodeFormatIds like this. You can get barcode format strings here.

    "Name": "task-read-barcodes-read-rate",
    "ExpectedBarcodesCount": 0,
    "BarcodeFormatIds": [
   -  "BF_DEFAULT"
   +  "BF_QR_CODE",
   +  "BF_DATAMATRIX"
    ],

4. Apply this template.

   CaptureVisionRouter.initSettings('path/to/the/template/file');

Warning

Due to its powerful customization capabilities, the number of configurable parameters in the templates is extensive and relatively complex. To ease your workload, if the preset templates do not meet your requirements, feel free to contact us for a customized template.

About the capture like API

captureAsync(...) processes images in worker. The maximum number of workers is defined by CaptureVisionRouter.maxWorkerCount, which defaults to <logical processor number> - 1 (minimum of 1). If you continue to call captureAsync(...) while all workers are busy, the tasks will be queued and wait for execution. You can get the queue length by CaptureVisionRouter.waitQueueLength.

The synchronous version is capture(...), which processes images on the main thread.

The capture -like APIs can accept file path string or file bytes Uint8Array as input data. Currently supported file types are jpg, png, bmp, gif, pdf, tiff. The capture like APIs also accept DCVImageData as input data. Typically used to process raw data from a camera.

interface DCVImageData{
  bytes: Uint8Array;
  width: number;
  height: number;
  stride: number;
  format: EnumImagePixelFormat;
  /** EXIF orientation; 1 or undefined means no rotate. */
  orientation?: number;
}

If input data is file bytes or DCVImageData, by default, captureAsync(...) will transfer bytes into worker. Thus you can't access these bytes in the main thread after captureAsync. This allows for optimal performance.

The following code can prevent bytes from being transferred.

let result = await CaptureVisionRouter.captureAsync(
  input_data_contains_bytes, {
    templateName: EnumPresetTemplate.XXXX,
    dataTransferType: 'copy'
  }
);

For multi-page PDFs and TIFFs, you can use captureMultiPages.

let results = await CaptureVisionRouter.captureMultiPagesAsync('./multi-page.pdf', EnumPresetTemplate.XXXX);
for(let result of results){
  const tag = result.originalImageTag;
  console.log(`# page ${tag.pageNumber}/${tag.totalPages}:`);
  for(let item of result.barcodeResultItems){
    console.log(item.text);
  }
}

Supported OS/Arch

Node.js >= 16.x

os	arch
win32 (windows)	x86, x64 (Vista or newer)
linux	x64 (glibc >= 2.18), arm64
darwin (mac)	x64, arm64

You can force(-f) install resource packages (dynamic libraries) for other OS/arch. So you can develop and deploy on different machines. You can check the <OS>-<arch>@<version> in this SDK's package.json->optionalDependencies.

npm i dynamsoft-capture-vision-for-node-lib-<OS>-<arch>@<version> -f -E

Samples

Web Service

Express sample and koa sample show how to use the SDK in a web service.

You do not need to start multiple process instances in PM2 Cluster mode. As mentioned above, Dynamsoft Capture Vision for Node already manages a thread pool. However, pm2 start app.js is still useful, it can automatically restart app.js when the service crashes.

AWS Lambda

We also made a special adaptation for AWS Lambda; see this sample. Other similar single-function platforms may have some compatibility issues. If you have any needs, please contact us.

PDF Advanced

If you want more control when processing PDFs, such as controlling the DPI or specifying which pages to process, please refer to this sample.

Special Notes

AI Model

When performing

• barcode recognition tasks excpet PT_READ_BARCODES_SPEED_FIRST
• text recognition tasks

you need to install dynamsoft-capture-vision-for-node-model. You can check the <version> in the SDK's package.json->peerDependencies.

npm i dynamsoft-capture-vision-for-node-model@<version> -E

Error Model file is not found

If you have installed the AI model package, and you called CaptureVisionRouter.initSettings(<custom barcode template>), then you still get Model file is not found error; you can fix it by adding a json to end of the template:

{
  "AAAA":[],
  "BBBB":[],
  "XXXX":[
+ ],
+ "CaptureVisionModelOptions": [
+     {
+         "Name": "Code128Decoder",
+         "MaxModelInstances": 4
+     },
+     {
+         "Name": "DataMatrixQRCodeLocalization",
+         "MaxModelInstances": 4
+     },
+     {
+         "Name": "EAN13Decoder",
+         "MaxModelInstances": 4
+     },
+     {
+         "Name": "OneDDeblur",
+         "MaxModelInstances": 4
+     },
+     {
+         "Name": "OneDLocalization",
+         "MaxModelInstances": 4
+     }
  ]
}

This is because in Node.js, the AI model location needs to be explicitly specified. The use of the model cannot be implicit; it must be explicitly stated so that its location can be modified by Node.js wrappers. The built-in templates already do this, but templates from other sources may not have considered the specific characteristics of the Node.js environment.

Multiple CaptureVisionRouter Instances

In some rare cases, you may need multiple CaptureVisionRouter instances, such as customizing different templates for each instance. Here is how to use:

let cvr = new CaptureVisionRouter();

let settings = cvr.outputSettings('<templateName>');
settings.foo = bar;
cvr.initSettings(settings);

let result = await cvr.captureAsync('<image>', '<templateName>');