How to avoid blocky reconstruction

[harshithabk]

@pathak22

How to remove the blocking artifact in the patch loction after reconstruction of image?
What value to be set in this part of code?

image_ctx[{{},{1},{1 + opt.fineSize/4 + opt.overlapPred, opt.fineSize/2 + opt.fineSize/4 - opt.overlapPred},{1 + opt.fineSize/4 + opt.overlapPred, opt.fineSize/2 + opt.fineSize/4 - opt.overlapPred}}] = 2117.0/255.0 - 1.0
image_ctx[{{},{2},{1 + opt.fineSize/4 + opt.overlapPred, opt.fineSize/2 + opt.fineSize/4 - opt.overlapPred},{1 + opt.fineSize/4 + opt.overlapPred, opt.fineSize/2 + opt.fineSize/4 - opt.overlapPred}}] = 2104.0/255.0 - 1.0
image_ctx[{{},{3},{1 + opt.fineSize/4 + opt.overlapPred, opt.fineSize/2 + opt.fineSize/4 - opt.overlapPred},{1 + opt.fineSize/4 + opt.overlapPred, opt.fineSize/2 + opt.fineSize/4 - opt.overlapPred}}] = 2*123.0/255.0 - 1.0

The above values are used to set mean value in R. G. B.
But we want to make this region fully transparent [no blocky artifact].

Can you please guide us on this regard

[pathak22]

What is a blocking artifact? Can you post some images?

[harshithabk]

@pathak22
Blocky artifact means

How do we remove those artifacts?

[pathak22]

Surprising! Is it the from the trained model I released? This very result looks much better when I tried. For instance see this figure here: https://people.eecs.berkeley.edu/~pathak/context_encoder/resources/result_fig.jpg

Could you give more details? thanks!

[russab0]

@harshithabk any details?
I have faced the same problem when trying to run the model on a different dataset. The reconstructed area is simply 16 blocks (attached the example)