Inappropriate content detection

[lvillacin]

Is your feature request related to a problem? Please describe.
Was hoping I could detect inappropriate content from the photos so I could blur them out unless I specifically don't want to.

Describe the feature you'd like
Maybe it could be one of the edits like an inappropriate filter option and the photos can come back blurred or even better somehow with a metatag that says it's inappropriate so I could do a custom handle in-app.

Either way, I think this is really useful for apps that allow user-generated content, which is, of course, one of the main uses for serverless-image-handler.

Additional context
I know this could be done using Amazon Rekognition and since serverless-image-handler already uses it for smartCrop, this might not be too hard to do I guess.

[beomseoklee]

@lvillacin thanks for your input. We put this one into our backlog item, and we will consider how to add this feature.

[lvillacin]

Thanks @beomseoklee looking forward to seeing this being implemented 🤩

[G-Lenz]

We've added support for content moderation in v5.2.0.

[jbaudanza]

We've added support for content moderation in v5.2.0.

This is a great feature! But would it be possible to have the contentModeration edit applied before the other edits in applyEdits.

The problem I have some users avatars that are being incorrectly flagged as inappropriate, but only the thumbnail versions. The original versions pass the content-moderation filter just fine.

[G-Lenz]

Hi @jbaudanza , the solution evaluates the edits in order, so if contentModeration is listed before the other edits, it should apply the contentModeration before the other edits.

[jbaudanza]

Hi @jbaudanza , the solution evaluates the edits in order, so if contentModeration is listed before the other edits, it should apply the contentModeration before the other edits.

Ah that works! Thank you!