LiuHong comments

Results 5 comments of


                                            LiuHong

使用pytorch训练出模型后转ONNX测试没问题，但是转成MNN后，精度严重下降

调一下nms

请问如何编译libMNN.so

我知道的是需要clone 下MNN配置cmake编译出对应安卓的.so文件发件人："sh" ***@***.***> 发送日期：2021-04-26 17:12:08 收件人：Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB ***@***.***> 抄送人：LiuHong ***@***.***>,Author ***@***.***> 主题：Re: [Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB] 请问如何编译libMNN.so (#249) 同问。readme里提供的哪个android demo libMNN.so跑不起来，会崩溃。目前也不知道怎么解决 mnn 需要根据官方MNN源码修改cmake 参数编译 — You are receiving this because you authored...

No module named 'torch._thnn'

i got the same problem

cannot import name 'Config' from 'mmcv'

Lite-HRNet$ python ./tools/train.py Traceback (most recent call last): File "./tools/train.py", line 9, in from mmcv import Config, DictAction ImportError: cannot import name 'Config' from 'mmcv' (/home/name/anaconda3/envs/lite-hrnet/lib/python3.8/site-packages/mmcv/__init__.py)

nvcc fatal : Unsupported gpu architecture 'compute_61'

my GPU is 1080ti,it's fit for 'compute_61'