Babyhamsta / Aimmy

Universal Second Eye for Gamers with Impairments (Universal AI Aim Aligner - ONNX/YOLOv8 - C#)
Other
807 stars 327 forks source link

Any ideas how to fix this ? #1426

Closed versaud5 closed 2 months ago

versaud5 commented 3 months ago

Use roboflow exported dataset . Used Ultralytics yolov8 to get .pt then converted it to onxx. Used 1 class "person"

Output shaped 1x84x8400 does not match the expected shape of 1x5x8400 This model will not work with Aimmy. Please use ONNX V8 model.

Babyhamsta commented 2 months ago

Onnx has different output shapes, unsure if this is a newer YOLO model or what but I've seen it before.

ghost commented 2 months ago

Onnx has different output shapes, unsure if this is a newer YOLO model or what but I've seen it before.

I think Im pretty sure he changed some of the default parameters that yolov8 used to train onnx to give to aimmy Im using different versions of yolov8.pt also with modify at the same time, havent cause this issue if you have the right settings for trainning