Tencent Cloud ID Card OCR

Problem Scenario

When a system needs to extract fields such as name, ID number, and validity period from a mainland China ID card image, while also detecting risks like photocopy, reshoot, PS traces, or temporary ID, handling OCR, cropping, and warnings separately can create duplicated logic. @tencent-adm/tencentcloud-ocr-idcard targets this development scenario by wrapping the Tencent Cloud OCR ID card API into a single skill.

How It Works

The skill runs through scripts/main.py and supports two image input modes: ImageBase64 and ImageUrl. You can specify CardSide as FRONT or BACK; if omitted, the service detects the side automatically. Core capabilities include:
- Front side: name, gender, ethnicity, date of birth, address, and citizen ID number
- Back side: issuing authority and validity period
- Additional processing: ID card photo cropping or portrait cropping
- Risk warnings: photocopy, reshoot, PS traces, temporary ID, reflection, invalid date, and other checks

The Config request parameter is a JSON string used to enable warning and processing switches. CardWarnType can select Basic or Advanced warning algorithm levels. If WarnInfos is needed, at least one warning switch must be explicitly enabled in Config.

Boundaries and Notes

The skill requires Tencent Cloud API credentials, including TENCENTCLOUD_SECRET_ID and TENCENTCLOUD_SECRET_KEY. Base64 image input is limited to 10MB. Warning detection is not fully enabled by default. When a user requests checks such as “detect PS” or “detect reshoot”, the caller should confirm the relevant Config switches before relying on warning output. The JSON result is suitable for form prefill or review-assist workflows, but warning codes should not replace a complete compliance review.