Introduction

In the plugin ecosystem of DeepSeek Harness (DSH), processing image data is a common requirement. xby-pic is a toolset plugin focused on visual tasks, aiming to provide developers with unified image recognition and detection capabilities without deploying heavy local models. Various visual functions can be implemented directly through API calls.

Plugin Overview

The plugin is maintained by xby-skill and is under the MIT License. It implements its features through external APIs and does not depend on local models, making it suitable for scenarios that require rapid integration of OCR, identity document recognition, and object detection capabilities.

Core Features

The plugin provides the following verified functional interfaces:

  • API Key Configuration: set_xby_apikey, used to set the API key. It supports automatic persistence and remains valid after restart.
  • General Text Recognition: ocr, ocr_pro, and ocr_handwriting, supporting recognition of documents, screenshots, and handwritten content.
  • Identity Document Recognition: plate_recognition (license plate), ocr_bank_card (bank card), ocr_biz_license (business license), ocr_driver_license (driver’s license), ocr_vehicle_license (vehicle registration), ocr_id_card (ID card), ocr_pass (pass), and ocr_passport (passport).
  • Object Detection and Recognition: daily_object_detection, cellphone_detection, ebike_detection, fire_detection, gesture_detection, head_person_detection, smoking_detection, wild_animals_detection, bird_detection, helmet_head_person_detection, and detect_pedestrian.
  • Biological and Scene Recognition: insect_recognition, plant_recognition, animal_recognition, pet_emotion_recognition, dish_recognition, and general_recognition.

Installation and Activation

The plugin is available in the DeepSeek Harness directory. Visit the Plugin Directory or check the GitHub repository for the source code and installation instructions.

Usage Instructions

Before using the plugin, you need to call set_xby_apikey to set the API key. When calling recognition or detection functions, the following three input methods are supported:
* Image file links
* BASE64 encoding
* File paths

Notes

  • Input Requirements: Some functions (such as ocr) require the input to be an image file link.
  • Permission Execution: The plugin runs with the permissions of the current DSH process.
  • Source Code Review: Before installation, please review the source code and license (MIT) on your own.

Conclusion

xby-pic provides rich processing capabilities for visual tasks. For more details, please refer to GitHub.