Skip to content

Latest commit

 

History

History
executable file
·
127 lines (96 loc) · 3.45 KB

README.md

File metadata and controls

executable file
·
127 lines (96 loc) · 3.45 KB
Raw

Barcode Reading Benchmark

A barcode reading benchmark tool based on Quasar + Vue. It can create barcode reading benchmark projects which save the test results and images in indexedDB. The benchmark can run purely in the browser.

The following SDKs are included for testing:

Extra SDKs are supported via HTTP using Barcode-Reader-Aggregator.

Online demo

Features

  • Run batch barcode decoding on an image dataset
  • Calculate metrics like reading rate, precision and average time
  • Compare the performance of different barcode reading libraries
  • Compare the performance in categories
  • Export and import of benchmark projects

How the Metrics are Calculated

  • Reading rate: correctly detected barcodes / total barcodes
  • Precision: incorrectly detected barcodes / all detected barcodes
  • Average time: total time elapsed / total files

How to Infer that a Barcode is Correctly Detected

  1. Check if there is a ground truth barcode which overlaps with the detected barcode using overlapped percent. (the area of the intersection / the area of the ground truth or the detected result which has a smaller area).
  2. Check if the barcode text is correct. If the barcode content is in raw bytes, compare the binaries. Since 2D barcodes use Reed–Solomon error correction, it is okay if some of the binaries do not match with the ground truth's (based on edit distance).

Dive into Formats

  • Barcode Detection Result

    export interface DetectionResult {
  elapsedTime:number;
  results:BarcodeResult[];
}

export interface BarcodeResult {
  barcodeFormat: string;
  barcodeText: string;
  barcodeBytes: string;
  confidence?: number;
  x1:number;
  x2:number;
  x3:number;
  x4:number;
  y1:number;
  y2:number;
  y3:number;
  y4:number;
}

    Barcode detection result is saved in the above format in JSON with the .json extension.

  • Ground Truth

    export interface GroundTruth {
  text:string;
  x1:number;
  x2:number;
  x3:number;
  x4:number;
  y1:number;
  y2:number;
  y3:number;
  y4:number;
  attrib:Attrib;
  value_attrib:ValueAttrib;
}

export interface Attrib {
  Type:string;
}

export interface ValueAttrib {
  Mode?:string;
}

    Ground truth is saved in the above format in JSON with the .txt extension.

  • Remote Projects

    Put the files like the following example to the dataset folder as remote projects. Users can download them to check them out locally. The files can be retrieved using the export feature.

    │  projects.json
│
│─ Project
        detection_result_filenames.json
        project_manifest.json
        results.zip
        [P]ISBN_18_0010.jpg
        [P]ISBN_18_0077.jpg

How to run the app

Install the dependencies

yarn
# or
npm install

Start the app in development mode (hot-code reloading, error reporting, etc.)

quasar dev

Build the app for production

quasar build

Customize the configuration

See Configuring quasar.config.js.