Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.
6403 Discussions

ReLU operation is giving output as expected

idata
Employee
688 Views

I am trying to implement Max Pooling followed by ReLU, but i noticed my output is incorrect for the ReLU,

 

here is my relu.prototxt

 

name: "pooling_relu_net"

 

input: "data"

 

input_shape

 

{

 

dim: 1

 

dim: 1

 

dim: 1

 

dim: 5

 

}

 

layer {

 

name: "pool"

 

type: "Pooling"

 

bottom: "data"

 

top: "pool"

 

pooling_param {

 

pool: AVE

 

kernel_size: 1

 

stride: 1

 

}

 

}

 

layer {

 

name: "pool_relu"

 

type: "ReLU"

 

bottom: "pool"

 

top: "pool"

 

}

 

can some one help me… is this prototxt file correct?

0 Kudos
4 Replies
idata
Employee
424 Views

@murthysristi If you are referring to the "top" parameter of your pool_relu layer, I usually see the output layer's "top" value being set to itself. I ran a quick mvNCCheck on this model and I did not have any issues.

0 Kudos
idata
Employee
424 Views

Tome , Can you share prototxt file and the mvNCCheck results?

0 Kudos
idata
Employee
424 Views

I was using your prototxt file above as a base. Since you said you wanted MAX pooling followed by ReLu, I modified your prototxt file as such:

 

name: "max_pooling_relu_net" input: "data" input_shape { dim: 1 dim: 1 dim: 1 dim: 5 } layer { name: "pool" type: "Pooling" bottom: "data" top: "pool" pooling_param { pool: MAX kernel_size: 1 stride: 1 } } layer { name: "pool_relu" type: "ReLU" bottom: "pool" top: "pool_relu" }

 

Here are the mvNCCheck results:

 

mvNCCheck v02.00, Copyright @ Movidius Ltd 2016 ****** WARNING: using empty weights ****** USB: Transferring Data... USB: Myriad Execution Finished USB: Myriad Connection Closing. USB: Myriad Connection Closed. Result: (1, 1, 5) 1) 3 0.912 2) 4 0.8667 3) 2 0.0 4) 1 0.0 5) 0 0.0 Expected: (1, 1, 5) 1) 3 0.912 2) 4 0.8667 3) 2 0.0 4) 1 0.0 5) 0 0.0 ------------------------------------------------------------ Obtained values ------------------------------------------------------------ Obtained Min Pixel Accuracy: 0.0% (max allowed=2%), Pass Obtained Average Pixel Accuracy: 0.0% (max allowed=1%), Pass Obtained Percentage of wrong values: 0.0% (max allowed=0%), Pass Obtained Pixel-wise L2 error: 0.0% (max allowed=1%), Pass Obtained Global Sum Difference: 0.0 ------------------------------------------------------------
0 Kudos
idata
Employee
424 Views

Thank you, Let me check

0 Kudos
Reply