A comprehensive computer vision tool for identifying vehicles, animals, and plants, alongside high-accuracy OCR for text and table extraction.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install vision-recognition-ocr
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install vision-recognition-ocr using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Vision Recognition + OCR?
Vision Recognition + OCR is a multi-purpose computer vision extension designed for the Openclaw Skills ecosystem. It provides AI agents with the ability to perceive and interpret visual data through Baidu Vision APIs, bridging the gap between raw images and structured information. This skill excels at both classification tasks—such as identifying specific car models or plant species—and text extraction tasks, making it an essential tool for developers building agents that handle physical documents or visual media.
By supporting local file paths, public URLs, and base64 encoded inputs, this skill ensures flexibility across various deployment environments. Whether you are processing screenshots for automated workflows or digitizing handwritten notes, this addition to your Openclaw Skills library provides the technical foundation for robust visual intelligence.
Vision Recognition + OCR Use Cases
- Identifying vehicle makes and models from street photos or dashcam footage.
- Extracting text from screenshots, digital invoices, and receipts for automated accounting.
- Digitizing handwritten documents and notes into machine-readable text formats.
- Converting image-based tables into structured data for analysis and reporting.
- Classifying biological entities like animals and plants for research or inventory purposes.
How Vision Recognition + OCR Works
- The skill accepts an image input via local path, public URL, or base64 string.
- The AI agent selects the specialized script required for the task, such as car recognition or table OCR.
- The skill authenticates with Baidu Cloud services using provided environment variables (Bearer tokens or API keys).
- The image data is processed by the relevant Baidu Vision API endpoint.
- Structured JSON results are returned to the agent, containing classification labels or extracted text blocks.
Vision Recognition + OCR Setup
To begin using this skill within the Openclaw Skills framework, ensure you have Python 3.11+ and configure your credentials:
# Set your Baidu API credentials
export BAIDU_API_KEY="your_api_key"
export BAIDU_SECRET_KEY="your_secret_key"
# Or use a Bearer Token for priority access
export BAIDU_BCE_BEARER_TOKEN="your_token_here"
Test the installation with a basic OCR command:
py -3.11 scripts/ocr_general_basic.py '{"url":"https://baidu-ai.bj.bcebos.com/ocr/general.png"}'
Vision Recognition + OCR Data Schema & Taxonomy
The skill utilizes a standard input schema for all vision tasks to maintain consistency across the Openclaw Skills ecosystem:
| Field | Type | Description |
|---|---|---|
| image_path | string | Path to a local image file (optional) |
| image_base64 | string | Base64 encoded image data (optional) |
| url | string | Publicly accessible image URL (optional) |
| top_num | integer | Number of classification results to return (1-20) |
| detect_direction | boolean | Detect and adjust for text orientation in OCR |
| paragraph | boolean | Consolidate OCR results into logical paragraphs |
Vision Recognition + OCR Advanced Features
- High-accuracy OCR mode for processing low-quality or complex documents.
- Specialized handwriting recognition with adjustable granularity at the word or letter level.
- Table recognition with the ability to export contents directly toward structured formats.
- Multi-directional text recognition support for images with non-standard orientations.
- Integration hooks for automotive data workflows including VIN and EPC lookups.
SKILL.md
Loading
Related Openclaw Skills
Reveal.js Slide Creator
An automated presentation engine that generates self-contained, interactive HTML5 slidedecks using the Reveal.js framework via CDN.
Market Events
A professional-grade financial tracking skill that reports upcoming or historical corporate events for a customizable watchlist of stock tickers.
Prediction
A professional-grade forecasting framework that turns vague guesses into auditable, numeric probabilities using reference classes and rigorous scorekeeping.
K-12 Educational SVG Animation Engineer
An AI-powered technical expert that transforms complex K-12 educational concepts into intuitive, flat-vector SVG animations delivered as standalone HTML files.
CogDx - Cognitive Diagnostics
CogDx provides external cognitive verification for AI agents to detect biases, audit reasoning, and ensure mathematical calibration of confidence levels.
Safe Exec
A specialized security skill designed to analyze script intent and detect malicious behavior before code execution.
Featured*
