Closed malharjajoo closed 6 years ago
I'm going to say no, simply because I can't remember what I chose for the final assessment.
The test input generation is always quantitatively/statistically similar to the examples though,
so the only change would be a small fuzzing amount (e.g. width=height+(rand()%16)
or
similar). There wouldn't be anything where width=10 and height=10000000
Ok, I asked this since if width = height, then the coefficient kernels would be completely identical.( which could save some time )
Hi,
@m8pple , Can we assume the width = height = scale always for gaussian_blur ?
From the CreateInput() in the reference file it seems to be the case, but would this be the same during testing ?