Custom yolov8 model with SAHI #969
-
I have trained a model on visdrone data using pre-trained weights for yolov8 . It is giving a map 0f around 30 . is it a good score ? . Also my next plan of action is to integrate the model I've built with SAHI because most of the objects in visdrone dataset are small . Thank you !!😊 |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments
-
@akashsateesha Have you got any results? I'm struggling with the same task. |
Beta Was this translation helpful? Give feedback.
-
Hey @akashsateesha @ablyzniuk, you can use your custom finetuned yolov8 models with sahi as given here: https://github.com/obss/sahi/blob/main/demo/inference_for_yolov8.ipynb Moreover, you can follow this example for sahi + yolov8 + object tracking 🚀 ultralytics/ultralytics#8642 |
Beta Was this translation helpful? Give feedback.
Hey @akashsateesha @ablyzniuk, you can use your custom finetuned yolov8 models with sahi as given here: https://github.com/obss/sahi/blob/main/demo/inference_for_yolov8.ipynb
Moreover, you can follow this example for sahi + yolov8 + object tracking 🚀 ultralytics/ultralytics#8642