Yolov8 Object Detect Bounding box messed

[Dieguinho5]

Hello friends,
I'm having a problem with my custom yolov8 template.
The bounding boxes are messed up and it's giving a pretty low prediction.
I'm using the starter project https://github.com/pytorch/android-demo-app/tree/master/ObjectDetection

Can anybody help me?

model yolov8.torchscript: https://drive.google.com/file/d/1-0IEYXOTyLkLlfrU6LeHy2SqJ7hh1g6G/view?usp=sharing

[janimo]

In general yolo v5 and v8 have differently shaped outputs so dropping in a v8 model is not supposed to work out of the box.

[encouver]

In general yolo v5 and v8 have differently shaped outputs so dropping in a v8 model is not supposed to work out of the box.

Do you have some reference to use with yolov8?

[NeighborhoodCoding]

Can you please share the full code on how to change the yolov5 code to yolov8 code?