每天监控PRADO和220多个政府登记册。将每个新版本与认证数据库进行比对。
涵盖 220 多个国家/地区的 14,000 多种证件。
搜索所有国家/地区,所有证件。
verification.didit.me/v1/organization/documents-allowed-by-country , 与 Didit 业务控制台读取的数据源相同。可按国家名称或 ISO 3166-1 alpha-3 代码搜索。研究人员每天查阅所有政府登记册。
自动将低分辨率缩略图添加到分类器。在高保真模板发布前,进行人工合规审查。
无需客户重新部署即可发布到生产环境。您的工作流即时接受新版本, 相同的API、相同的JSON结构、相同的价格。
常见问题
How do you add a brand-new document type you have never seen?
Research agents monitor PRADO (the EU Public Register of Authentic Documents Online), the U.S. Department of State document library, and 220+ per-country government registries every day. When a new variant ships, a new German Personalausweis, a redesigned Brazilian CNH, a freshly-issued Estonian eID, the agent diffs it against the certified database, files the specimen, and routes it through one of two pipelines depending on capture quality.
No customer redeploy required, the new variant appears on the searchable table on this page the moment it ships.
What decides auto-add versus manual review?
Capture quality. Two paths:
- Auto-add via PRADO, low-resolution variants (thumbnails under 600 px on the long edge, the size PRADO publishes for most documents) flow straight into the classifier. They teach the model to recognise the document on sight but are too small to read security features from.
- Manual review, high-fidelity specimens (full-resolution captures with visible Optical Character Recognition (OCR) fields, Machine-Readable Zone (MRZ) zones, and security features) go through compliance review before they ship.
Both paths land on the same /v3/ session API and return the same JSON shape, you do not see the difference downstream.
How do you handle expired documents?
Document Monitoring tracks every user's document expiry date after the initial verification. When a document expires, the user status flips automatically from Approved to Kyc Expired and the platform fires a webhook so your app can prompt re-verification.
The same monitoring layer powers ongoing Know Your Customer (KYC) for regulated workflows. Full reference at docs.didit.me/core-technology/id-verification/document-monitoring-id-verification.
How does NFC reading work, and which documents support it?
Near-Field Communication (NFC) reads the cryptographic chip embedded in eMRTDs, Electronic Machine-Readable Travel Documents, the ICAO 9303 standard. That covers virtually every modern e-passport plus EU national eID cards and EU residence permits.
The phone reads the chip wirelessly, verifies the Document Signer Certificate against the issuing country's Country Signing Certificate Authority (CSCA), and returns a tamper-proof copy of the document data. NFC ships in the iOS and Android native Software Development Kits (SDKs), browser-based flows fall back to standard ID Verification + selfie because the Web NFC API is unstable on Android and absent on iOS.
How accurately do you parse the Machine-Readable Zone (MRZ)?
The Machine-Readable Zone (MRZ) is the two- or three-line code at the bottom of every ICAO 9303 passport and most EU national ID cards.
Didit parses it with full check-digit validation, every field (document number, date of birth, date of expiry, nationality, document type) is mathematically validated against its check digit before the response returns. The MRZ also acts as a structured fallback whenever Optical Character Recognition (OCR) confidence drops below threshold, so name, document number, and dates always have a deterministic recovery path even on imperfect captures.
Do you verify document authenticity, security features, and holograms?
Yes. Every session runs:
- Template matching against the certified specimen.
- Security-feature validation, holograms, watermarks, Optically Variable Ink (OVI), kinegrams, microprint.
- Document liveness, defeats screen replays, printed copies, and photographs of screens.
- Portrait tamper detection.
- Machine-Readable Zone (MRZ) check-digit validation.
- Country Signing Certificate Authority (CSCA) chain check when Near-Field Communication (NFC) is available.
Full catalog of warnings, what each one means and how it surfaces, at docs.didit.me/core-technology/id-verification/warnings-id-verification.
How fast does document inference complete?
status.didit.me.How do you handle low-quality or blurry captures?
Every side of the document carries a front_image_quality_score and back_image_quality_score in the report, each scoring 0-100 across focus, glare, occlusion, and crop.
The intelligent-capture flow surfaces real-time visual cues during capture (positioning, lighting, focus) and only auto-captures when conditions are ideal, so most low-quality submissions never reach the model. When a captured image still scores low, the session emits a warning and either auto-retries the user or routes to manual review, depending on your workflow.
What is the difference between ID Verification and Database Validation?
- ID Verification ($0.15) reads the document itself, Optical Character Recognition (OCR), Machine-Readable Zone (MRZ), Near-Field Communication (NFC), security features, document liveness. It answers is this a real document, and what does it say?
- Database Validation (from $0.05) takes the data ID Verification extracted and queries it against the authoritative government registry that issued the document, RENAPER (Argentina), Receita Federal (Brazil), Tribunal Electoral (Panama), RENAPO (Mexico), and 60+ others. It answers does that data actually match the official record, and is the photo on file the same person as the selfie?
Run them together for the strongest defence against synthetic-identity fraud.
What does a document check cost?
Run ID Verification standalone at $0.15. Bundle it into the recommended $0.33 KYC (Know Your Customer) workflow with ID Verification ($0.15) + Passive Liveness ($0.10) + Face Match 1:1 ($0.05) + Device & IP Analysis ($0.03) on a single API call.
Get 500 verifications free every month, every workspace, forever, no credit card required. No minimum commit, no overage surprise: the price on didit.me/pricing is the price on the invoice.
How do I request a document or a language you do not yet support?
Contact us on WhatsApp at https://wa.me/+19544659728 or ping the team from the Business Console. We ship a new document, or a new language, to production in 1 day when you request it. The research agent pulls every available specimen, the model retrains, manual-review compliance signs off, and the new variant lands on the same /v3/ API with no customer redeploy on your side.
The verification flow already ships in 49 languages out of the box. Need one we do not have? Ask, same 1-day turnaround, with full localised capture, retry prompts, and error states from day one.