tencentcloud-ocr-licenseplate_skill

Overview

This skill calls Tencent Cloud LicensePlateOCR to automatically locate and recognize mainland China vehicle license plates. It returns plate number, color, confidence, and pixel coordinates, and supports multiple plates per image and both Base64 and URL inputs. It is optimized for quick integration in Python environments using the official SDK.

How this skill works

The skill sends an image (Base64 or URL) to the Tencent Cloud OCR endpoint and parses the JSON response. It extracts Number, Confidence, Color, LicensePlateCategory, Rect and a LicensePlateInfos list for multi-plate results. Default request throttling is 10 requests/second and the SDK requires Tencent Cloud credentials in environment variables.

When to use it

• You need to extract vehicle plate numbers from photos or CCTV frames of mainland China vehicles.
• You require plate color and confidence metrics for validation or filtering.
• Images may contain multiple vehicles and you want all detected plates returned.
• You need pixel coordinates to crop or track plates within the original image.
• You prefer a managed OCR API with Python SDK support and simple inputs (URL or Base64).

Best practices

• Provide either ImageUrl or ImageBase64 (both allowed but URL takes priority) and keep encoded size under 10 MB.
• Use clear, reasonably sized images; avoid GIFs and ensure download time for URLs is within 3 seconds.
• Respect the 10 requests/second default limit; implement retry/backoff on throttling or transient errors.
• Validate Confidence values and apply business rules (e.g., thresholding) before trusting results.
• Handle error codes explicitly (download, decode, OCR failure, billing or quota issues) for robust flows.

Example use cases

• Parking lot gate: detect and log incoming/outgoing plate numbers with timestamps and crop rectangles.
• Toll or access control: validate plate color/type and match against whitelist/blacklist.
• Traffic monitoring: extract multiple plates per frame for vehicle counting and incident review.
• Evidence collection: save plate crops plus Confidence and RequestId for audit trails.
• Fleet management: automatically register vehicle arrivals from mobile or fixed cameras.