xby-gesture-detection

Recognizing hand gestures in an image and turning the result into structured data -- bounding boxes, confidence scores, class labels -- typically means pulling in a model, writing inference glue code, and babysitting input/output formats. xby-gesture-detection collapses that whole pipeline into a single conversational call inside DeepSeek Harness, so you skip the local deployment and boilerplate entirely.

The plugin accepts the image in whichever form is most convenient: a file URL, a Base64-encoded string, or a local file path. A single call returns every gesture target found in the frame, each annotated with pixel-level box coordinates, a confidence score, and a category label ready for downstream labeling, moderation, or multimodal interaction logic.

Whether you are building a vision-understanding workflow in DSH, prototyping a multimodal Q&A demo, or simply want to check the gesture distribution across a batch of screenshots without writing any code, this plugin lets you focus on the business decision rather than the inference plumbing.