In the development workflow of DeepSeek Harness (DSH), adding visual recognition capabilities to an agent usually requires introducing an external inference service. Integrating this capability directly at the plugin layer can reduce intermediate service calls and simplify the pipeline. The xby-daily-object-detection plugin provides this capability.

Functional Positioning

This is a DSH model inference plugin. Given an image, the plugin detects people, pets, vehicles, flames, or cardboard boxes within it and outputs the detection boxes, confidence scores, and labels for detected objects.

Core Capabilities

The plugin provides the following functions:

  • set_xby_apikey: Sets the API key in the chat, with automatic persistence and remains effective after restart.
  • daily_object_detection: Accepts an image file link and performs detection.
  • daily_object_detection_for_data_base64: Accepts the Base64 encoding of an image file and performs detection.
  • daily_object_detection_for_data_file: Accepts the file path of an image file and performs detection.

Installation and Activation

Install the plugin using the following command:

dsh plugin --profile web add github:xby_skill/xby-daily-object-detection

After installation, the plugin runs with the permissions of the current DSH process. The plugin is open source under the MIT License.

Configuration and Prerequisites

Before using it, obtain an API key from the Xiaobenyang official website.

Typical Usage

Depending on the input format, call the corresponding function:

  1. Input image file link: Call daily_object_detection, with the image URL as the parameter.
  2. Input image file Base64 encoding: Call daily_object_detection_for_data_base64, with the Base64 string as the parameter.
  3. Input image file path: Call daily_object_detection_for_data_file, with the local file path as the parameter.

Notes

  • This plugin is maintained by community developer xby-skill and has no official affiliation with DeepSeek or Huanfang.
  • It is recommended to check the source code before installation to ensure it meets project requirements.

By encapsulating external model capabilities, this plugin provides DSH with a convenient object detection solution. The project source code and more details are available on GitHub.