We were unable to load Disqus. If you are a moderator please see our troubleshooting guide.

NSFW

Oof

thank you!

Interesting stuff, thank you! Just one remark: the term "convolution" implies the fact that the kernel matrix is "flipped" (flip the elements order in the rows, flip the elements order in the columns). You are just multiplying the corresponding values from the image and the kernel, without flipping them. I think the implementations are not correct, since not flipping the kernel implies another operation called cross-correlation, which is not equivalent to the convolution. Indeed, being a symmetric matrix, there's no reason to flip, but I think that's a detail worth mentioning.

this is so complicated #confusedteach mehow

wow

cup_of_tea• 4 years agoWhy using the cropped original image? :^)