HaloAnnotator does not work

[wilsonlv]

Search before asking

• I have searched the Supervision issues and found no similar feature requests.

Question

can you give me a complete code? when i use HaloAnnotator, nothing changes in the image, the detections.mask is None

import cv2
import supervision as sv
from ultralytics import YOLO

filename = '../src/static/dog.png'
image = cv2.imread(filename)

model = YOLO('../models/yolov8x.pt')
results = model(image)[0]

detections = sv.Detections.from_ultralytics(results)

annotated_image = sv.HaloAnnotator().annotate(
    scene=image.copy(), detections=detections)

sv.plot_image(annotated_image)

Additional

No response

[LinasKo]

Hi @wilsonlv 👋

It's very useful that you checked the mask value.

To use HaloAnnotator you need a segmentation model. For example:

model = YOLO('yolov8x-seg')

[wilsonlv]

@LinasKo appreciate for help!

[SkalskiP]

@wilsonlv we also mention the detections.mask requirement in docs: https://supervision.roboflow.com/latest/detection/annotators/#supervision.annotators.core.HaloAnnotator :)