Hey,
No concrete estimates yet. We will share an update when tools.luxonis.com are updated.
Best,
Matija
YOLOv8
22 days later
Hey @ChrisCoutureDelValle , circling back to say that we now support YoloV8 in tools. We are working on improving the decoding on the FW, which will likely come out in the next DepthAI release. Regardless, you should be able to use YoloV8 with the latest existing DepthAI version.
Matija Awesome, thanks so much!
I tried with yolov8m pretrained model,and running examples/Yolo/Tiny_yolo script by modifying nnPath and camRgb.setPreviewSize(640, 640).Also removed achors and setachormasks as yolov8 is a achor free model.But detections are not happening correctly.
cc Matija on above ^. Otherwise @susantini please submit an Issue on github; depthai-ml-training repo.
Thanks, Erik
Hey @susantini , thanks for reporting the issue.
I'd suggest you try out an example from here. You should be able to run it with:
python3 main_api.py -m <model_name> --config <config_json>where model is the blob file and config the JSON file you get from tools.luxonis.com. If that still produces the same results, please share the .pt weights and an example input image, and we will further look into it.
I am having an issue as well with the converted yolov8 model. 
8 days later
Commenting to keep this thread active @Matija @erik @susantini
ChrisCoutureDelValle For trackability purposes, could you create issue on depthai-ml-training repo?
@ChrisCoutureDelValle Have you tried using the example I listed above? If yes, please share some more details:
- DepthAI version (latest DepthAI version is required for YoloV8 to work correctly)
- Script you're using to run the model
2 months later
Worked for me. Converted the model as suggested via the https://tools.luxonis.com/ and successfully ran it. Grand!
