转TensorRT

https://github.com/NVIDIA-AI-IOT/torch2trt

  

   

    

     

    
    

     

      pip3 install tensorrt-5.0.2.6-py2.py3-none-any.whl  
     
    
   

    

     

    
    

     

      # pycuda
     
    
   

    

     

    
    

     

      pip3 install pycuda 
     
    
  
 

按照这个说法，tensorrt提升不到2倍。

config:

https://github.com/tensorboy/centerpose/blob/master/experiments/mobilenetv2_512x512.yaml

https://github.com/bleakie/CenterMulti 

人体17个关键点，人脸是5个关键点

self.hps = nn.Sequential( nn.Conv2d(intermediate_channel, head_conv, kernel_size=3, padding=1, bias=True), nn.ReLU(inplace=True), nn.Conv2d(head_conv, 10, kernel_size=1, stride=1, padding=0))

文章来源: blog.csdn.net，作者：网奇，版权归原作者所有，如需转载，请联系作者。

原文链接：blog.csdn.net/jacke121/article/details/104543467