Does DJL support YOLOv8-based object detection models trained on custom datasets?

Open geekwenjie opened this issue 5 months ago • 3 comments

Jul 29 '25 03:07 geekwenjie

@geekwenjie

Yes, you can export your own Yolov8 model to torchscript or Onnxruntime, the DJL can load the model and use Yolov8 translator to run inference

Jul 29 '25 04:07 frankfliu

@frankfliu I know that DJL can be used for YOLO inference, but does it support training directly on DJL?

Jul 29 '25 05:07 geekwenjie

No, we don't support training YOLO model in Java. It's better you train the model in Python and then load it in DJL for inference.

Jul 29 '25 14:07 frankfliu