search

twlelev / Yolov9-Ncnn

ncnn deploy
16 stars 4 forks source link

ncnn Models #1

Open Qengineering opened 8 months ago

Qengineering commented 8 months ago

Where can we download yolov9-c-converted-sim.param and yolov9-c-converted-sim.bin ?

twlelev commented 8 months ago

I apologize for only seeing the message now, I have already uploaded the model.

aaafdsf commented 8 months ago

怎么转换到onnx再转换成ncnn模型呢

aaafdsf commented 8 months ago

Later on, I will provide instructions on how to convert it to a model supported by ncnn. 教程什么时候出

stackByStack commented 8 months ago

Later on, I will provide instructions on how to convert it to a model supported by ncnn. 教程什么时候出

You can reference this url. https://github.com/DataXujing/YOLOv9/tree/main/ncnn

aaafdsf commented 8 months ago

yolov9-c.onnx经过get_ncnn_onnx.py后得到last_ncnn.onnx，再通过onnx2ncnn 转换时仍然报错 Unsupported unsqueeze axes ! Unsupported unsqueeze axes ! Unsupported unsqueeze axes ! Unsupported unsqueeze axes ! Unsupported unsqueeze axes ! Unsupported unsqueeze axes ! Unsupported unsqueeze axes ! Unsupported reduction axes ! Unsupported unsqueeze axes ! Unsupported reduction axes ! Unsupported unsqueeze axes ! Unsupported reduction axes !

stackByStack commented 8 months ago

用export.py产生的文件，opset设为12（我自己的经验），加--simplify(大概是这个）

aaafdsf commented 8 months ago

python export.py --weights yolov9-c.pt --include onnx --simplify 转换命令应该是没有问题 可能是我运气不好

teohjianye commented 2 months ago

python export.py --weights yolov9-c.pt --include onnx --simplify 转换命令应该是没有问题

可能是我运气不好

Hi do you manage to convert your own pt model to ncnn and run on Android eventually?