[ndm000]

Shameless plug - I've been working on something similiar, but with a different aim.

https://github.com/nmurray1984/porn-blocker-chrome-extension

I found that the hardest problem to solve was the prevanlence of false positives. Even if you have a low false positive rate, it's still very likely to have an image blocked regularly just due to volume.

For that reason, the focus of my plugin is - for users that are OK with it - contributing URLs that are not NSFW but have been blocked. The extension includes a right-click menu option to do so.

[antpls]

Assuming you meant some SFW images are mistakenly blocked, it is an easier situation than with true negative.

Once an image tested positive with your first model, you could run a second more CPU intensive model (but also more accurate) on it.

[wongarsu]

Assuming you have a notably more accurate algorithm. It's a difficult problem even with infinite computing power.

[ndm000]

Great idea. MobileNetv2 is the model used currently due to it's small footprint (I used transfer learning on a dataset of NSFW and SFW images). A more robust model ran on all positives could improve performance overal.