search

chshin10 / epinet

EPINET: A Fully-Convolutional Neural Network using Epipolar Geometry for Depth from Light Field Images
MIT License
90 stars 39 forks source link

More details in augmentation would be appreciated #6

Open ghost opened 6 years ago

ghost commented 6 years ago

Hello Changha Shin. In the paper, you mentioned that methods such as randomly converting color to gray scale from [0,1] is used. However, by reviewing the "epinet_fun/func_makeinput.py" I notice that it've already converted the image into gray scale before returning the data. Is the training data all gray or just part of it? Is this function only applied in the test process or in the training as well? More details in augmentation would be appreciated. Thank you for your great work.

chshin10 commented 6 years ago

Yes, training data is also all gray scale. We use a randomly gray-scale conversion. img_gray=random0to1_R*img_Red+ random0to1_G *img_Green+ random0to1_B *img_Blue (Constraint: random0to1_R +random0to1_G+random0to1_B = 1, min(img_gray)=0 , max(img_gray)=1 )

I found that gray-scale conversion improves overall performance quite significantly perhaps because it reduces the overall complexity while conserving the light field structure.

ghost commented 6 years ago

That'll explain it. Thank you!