Introduction¶
The core philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” In model inference scenarios, object detection typically provides only bounding boxes for objects, whereas instance segmentation goes further by precisely separating objects from the image background. xby-seg is designed to address this need by providing YOLO26-based instance segmentation capabilities.
Plugin Introduction¶
xby-seg is a model inference plugin maintained by developer xby-skill. It focuses on instance segmentation of 80 COCO object classes in images and outputs bounding boxes, masks, confidence scores, and class labels.
Core Features¶
The plugin provides the following functional modules:
- set_xby_apikey: Sets the API key in the conversation. This operation is automatically persisted and remains valid after restarting DSH.
- yolo26_segment: Performs instance segmentation. Takes a link to an image file and returns the segmentation results.
- yolo26_segment_for_data_base64: Performs instance segmentation. Takes the BASE64-encoded content of an image file and returns the segmentation results.
- yolo26_segment_for_data_file: Performs instance segmentation. Takes the file path of an image file and returns the segmentation results.
Installation and Enablement¶
Run the following command in the terminal to install:
dsh plugin --profile web add github:xby_skill/xby-seg
Configuration and Usage¶
Before using this plugin, you must obtain an API key. Visit the Xiaobenyang official website to register and obtain the key.
A typical usage workflow is as follows:
-
Set the API key:
Invokeset_xby_apikeyin the chat and enter the key you obtained. -
Perform instance segmentation:
Depending on the input data source, choose one of the following three commands:- When using
yolo26_segment, provide the image file link. - When using
yolo26_segment_for_data_base64, provide the BASE64-encoded content of the image file. - When using
yolo26_segment_for_data_file, provide the image file path.
- When using
Notes¶
- The plugin is licensed under the MIT license.
- The plugin runs with the permissions of the current DSH process. It is recommended to review the source code and license before installation.
- The plugin depends on an external API service. Ensure that your network connection is working properly.