@Gargron given that the spam is mainly the same images, could you hash them and use that as a rejection filter?
Top-level
@Gargron given that the spam is mainly the same images, could you hash them and use that as a rejection filter? 1 comment
|
@KevinMarks @Gargron Assuming they use the exact same image, possibly. But if they even so much as slightly change the image (e.g., convert to another format, change some colour mapping etc) then it won't work with traditional hashing.
There exist hashing methods that work on visual similarity, but those are more complicated, and significantly harder to get right.
Also, more vulnerable to false positives, and worse catch rate.