High Inference Time with YOLO model

[DFerrari13]

Hi everyone,
I'm developing a flutter application for pose detection on a custom dataset using YOLOv11-pose by Ultralytics.
I noticed that, using a YOLO small model with input size of 128 pixels fine-tuned to detect an object with 9 keypoints, the inference time is around 150-200ms. According to the results provided on the website of Ultralytics I would have expected way lower inference time. Has anyone experienced the same?
Thanks
Davide