SDK de lecteur de code-barres

Scandit's Web SDK gets you to a first scan in under an hour (median time). You install two npm packages (@scandit/web-datacapture-core and @scandit/web-datacapture-barcode) or load them from a CDN, add your license key, and point the library at the camera. Pre-built components like the default scanning view and MatrixScan Find handle the camera, AR overlays, and loading states, so you are not writing low-level video code yourself. For an even faster route, install Scandit Agent Skills and let a coding agent (Claude Code, Codex, Cursor, GitHub Copilot, Gemini, and 40+ others) write the integration into your codebase, including setup, license activation, UI wiring, and performance defaults. See Agent Skills documentation.

In 2026, the easiest way to add barcode scanning to a website or web app with optimal performance is to combine an AI coding assistant with barcode scanning Agent Skills published by a recognized vendor of barcode scanner SDKs. You get the optimal solution for your use case without lengthy evaluation or customization — it just works within a few prompts.

Skills let your AI coding assistant execute pre-written, deterministic code, instead of generating and re-deriving a solution from scratch each time. Scandit Agent Skills are based on the Agent Skills Open Standard. They bring 15 years of barcode scanning expertise directly into any AI coding agent.

Scandit's Web SDK is a JavaScript library with proper TypeScript support: type definitions ship with the packages and every code example is written in TypeScript. You import only what you use, so tree-shaking keeps the bundle small. Because it is plain JS/TS with no framework lock-in, it fits any web stack. Scandit has accuracy levels browser scanners usually cannot reach: read rates consistently above 99%, reliable capture of tiny, torn, curved, and glare-hit barcodes in low light and motion, and an OCR fallback that reads the printed digits when a barcode is too damaged to decode. In one test it read a variety of common retail barcodes from an average distance of 77cm.

All of them. The Web SDK is a framework-agnostic JavaScript and TypeScript library, so it works with React, Angular, Vue, and Svelte the same way any npm package does. For frameworks that render on the server, like Next.js, Nuxt, and Gatsby, run the SDK on the client only and turn off server-side rendering for the scanning component; the Scandit docs cover this and link to Next.js's and Gatsby's own guides for disabling server-side rendering. (For React Native, use Scandit's dedicated React Native SDK.)

Yes. For a Progressive Web App, you can make scanning work fully offline by caching the SDK's .wasm and JavaScript files with a service worker; there are ready-made settings for Workbox and the Vite PWA plugin. One caveat: iOS Safari has a long-standing WebKit bug affecting camera access inside PWAs, so test on iOS. For Capacitor and Cordova, use Scandit's dedicated Capacitor SDK or Cordova SDK, rather than the raw Web SDK. You can also embed the Web SDK inside a native WebView using the WebView Integration guide. Ionic apps run on Capacitor or Cordova, so use those SDKs.

Entirely client-side, with no plugin or app to install. The Scandit Web SDK opens the camera through the standard MediaDevices.getUserMedia API, which needs a secure HTTPS context and the user's permission. Scanning runs in WebAssembly, offloaded to Web Workers so it does not block the main thread, with WebGL and OffscreenCanvas for GPU acceleration and WebAssembly SIMD to speed up OCR. All image processing happens on-device by default, so no images leave the browser.

Open-source libraries like ZXing, ZBar, and Google ML Kit are free and fine for clean, well-lit, one-at-a-time scans. They fall down on damaged, tiny, angled, or distant barcodes and on high-volume capture, and most come with no support or maintained roadmap (ZXing has been in maintenance mode since December 2025).

In Scandit's head-to-head tests, it read all 11 damaged "tough" labels where ZXing read 2, with a 0% false-positive rate against ZXing's 5% and a 168cm scan range against 37cm; against ML Kit it was up to 4X faster on shelf-label workflows. Scandit also offers enterprise support. Both Swiss Federal Railways and MG-DX chose the Scandit Web SDK over open-source. See our ZXing and ML Kit blogs for more information. (N.B. While the scanning engine is shared across Scandit's web and native SDKs, these tests were run on mobile devices, not the browser.)

Dynamsoft, Scanbot, STRICH, and Scandit are all commercial barcode SDKs, so compare them on scan performance, breadth, and support, not price alone. In Scandit's July 2026 head-to-head against Dynamsoft (Scandit 8.4.0 vs Dynamsoft 11.4.1200, on a mid-range Samsung Galaxy A56), Scandit scanned from a median 1.4X further away across seven barcode types, up to 2.7X further for bottom-shelf electronic shelf labels, and matched or beat Dynamsoft in 19 of 23 tough-code conditions. On a busy grocery shelf, Scandit captured the right label 7 out of 7 times in about 7 seconds, while Dynamsoft averaged 9 unintended scans per run and took 31 seconds. Scandit also adds context-aware AI, native AR overlays, and pre-built UI components that Dynamsoft either lacks or covers through third-party integration. The trial is self-serve. See Scandit vs Dynamsoft for full details. (Results from mobile-device testing, not the Web SDK; the scanning engine is shared across Scandit's web and native SDKs.)

For scanning many barcodes quickly, Scandit has MatrixScan Batch for high-throughput multi-code capture, plus MatrixScan Find and MatrixScan AR, which add on-screen overlays to guide picking, counting, and flagging. One press can capture every barcode in view. Using MatrixScan Batch, US mobility specialist Stratix were able to process shipments and deliveries 50% faster.

The Web SDK reads all major symbologies: 1D retail codes (e.g. EAN/UPC, Code 128, Code 39, Code 93, Interleaved 2 of 5), 2D codes (e.g. QR, Micro QR, Data Matrix, PDF417, Aztec, MaxiCode, DotCode), GS1 DataBar and composite codes, and postal codes (e.g. USPS Intelligent Mail, KIX, RM4SCC, UPU S18). Browser support covers Chrome and Edge 85+, Firefox 105+, Safari 15.2+ (14.6+ single-thread), Opera, Samsung Internet, and Chrome and Safari on iOS and Android. The camera needs an HTTPS context. Full symbology and browser-compatibility tables are in the docs.