torch 多维topk

flatten和view(-1) 是按行展开，一行一行的拼接

  

   

    

     

    
    

     
 import torch
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

       anch_ious = torch.Tensor([[1, 2, 3,4], [7,8,9,3]])
     
    
   

    

     

    
    

     
 # anch_ious = torch.Tensor([[1, 2, 3,4], [4, 5, 6,7]]).view(-1)
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

       neg_count=3
     
    
   

    

     

    
    

     

       top_data,index= torch.topk(anch_ious, neg_count, dim=1, largest=True, sorted=True, out=None)
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 # print(anch_ious[index])
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

       index_num = torch.arange(0, anch_ious.size(0)).repeat(3,1)
     
    
   

    

     

    
    

     

       index_num=torch.flatten(index_num.t())
     
    
   

    

     

    
    

     

       print(index_num)
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 # indexs = index[:, 1:]
     
    
   

    

     

    
    

     

       index = torch.flatten(index)
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

       print(anch_ious[index_num,index])
     
    
  
 

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/104820949