dingniaoji

@acracker 你的任务很有意思，希望不是用作`非法`用途。 这是我在`wsl2`中解决此问题的方案，供你参考。 ## 中文乱码（训练前都解决掉，不然从头来就麻烦了） ultralytics 8.2.91-100 ultralytics最近更新很频繁，每次更新我们都需要修改下`metrics.py`文件，其它不用修改 ### 在wls2中把windows的fonts直接给到ubuntu ``` ln -s /mnt/c/Windows/Fonts/ /root/.local/share/fonts ``` 1. 复制SimHei字体到特定目录（并非必须，因为上面windows共享的里面有） ``` /root/anaconda3/envs/yolov8/lib/python3.10/site-packages/matplotlib/mpl-data/fonts/ttf ``` 2. 修改`matplot配置` ``` vim /root/anaconda3/envs/yolov8/lib/python3.10/site-packages/matplotlib/mpl-data/matplotlibrc font.family : sans-serif #...

for others who came to this question: the problem is about the model arch you chose, m/x may not work at mobile.

I think if you can make the `transfer` method work as expected in `models` page, the problem will be solved. But it may be a busssiness concern or it is...

[here is the more detailed disscusion](https://github.com/ultralytics/hub/issues/833) I hope it is a `bug`, not a `bussiness concern`. :)

I have a similar use case to this. I have a `object detection model` for each platform. `tflite` for android. `mlmodel` for ios. both of them are very big.

> Sorry If I am a bit persistent, but I need to know if Ultralytics intends or doesn't intend not persue tflite micro as a supported use case for me...

@glenn-jocher if `yolo-flutter-app`(https://github.com/ultralytics/yolo-flutter-app) support `f16` and `f32`, `int8` problem is not that annoying for me . now I have to switch to other yolo mobile solution like `NCNN`.

> We understand the importance of INT8 support and are looking into it. Meanwhile, using FP16 or FP32 with the yolo-flutter-app is a viable option. Thank you for your patience....

> @rjacaac I haven't solved it yet but a small update, it works with yolov8: > > ```python > from ultralytics import YOLO > model = YOLO('yolov8n.pt') > model.export(format="tflite", int8=True,...