Add-on Proposal
Overview
We can register for a variety of add-ons that offer unique image and/or video processing or delivery capabilities. This proposal document is to cover the two add-ons we would like to enable on our account and their capabilities.
Phase 1
Amazon Rekognition Auto Tagging
Amazon Rekognition is a service that makes it easy to add image analysis.
Cloudinary provides an add-on for Amazon Rekognition's automatic image tagging capabilities, fully integrated into Cloudinary's image management and transformation pipeline. Amazon Rekognition uses deep neural network models to automatically identify scenes and suggests tags, a process that would take huge amounts of time and resources if performed manually. Currently this process is not performed at all and teams rely upon filenames. The issue here is that filenames do not follow a convention and are usually poorly named making it impossible to search for assets correctly.
Amazon Rekognition's deep learning-based image recognition allows us to search, verify and organize all our images and results in outstanding performance in terms of the time needed to process large photo volumes with a very high precision rate.
With the Amazon Rekognition Auto Tagging add-on, we can automatically extend Cloudinary's powerful semantic data extraction and image tagging features. When using the Amazon Rekognition Auto Tagging add-on, our images are automatically tagged according to the categories detected in each image.
Phase 2
Amazon Rekognition AI Moderation
With Amazon Rekognition's AI Moderation add-on, we can extend Cloudinary's powerful cloud-based image media library and delivery capabilities with automatic artificial intelligence based moderation of our photos. Protect our users from explicit and suggestive adult content in our user uploaded images, making sure that no offensive photos are displayed to our web and mobile viewers. The main driver here is to protect our DAM from UGC that is saved via the mobile application.
Automatic image modernation flow
The following list describes the flow of uploading and displaying moderated images using Cloudinary and the Amazon Rekognition's AI Moderation add-on:
- Our users upload an image to Cloudinary through our mobile application.
- The uploaded image is sent to Amazon Rekognition for moderation.
- The image is either approved or rejected by Amazon Rekognition.
- An optional notification callback is sent to our application with the image moderation result.
- A rejected image does not appear in our media library, but is backed up, consuming storage, so that it can be restored if necessary.
- Moderated images can be listed programmatically using Cloudinary's Admin API or interactively using our online Media Library Web interface.
- We can manually override the automatic moderation using the Admin API or the Media Library.
Moderation categorization
Amazon Rekognition assigns a moderation confidence score (0 - 100) indicating the chances that an image belongs to an offensive content category.
There are two levels of categories for labelling unsafe content, with each top-level category containing a number of second-level categories, for example under the 'Violence' (violence) category you have the sub-category 'Physical Violence' (physical_violence). A full list of all the latest available categories and sub-categories is provided by AWS, see Amazon Rekognition categories in the resources section.
The top level categories include:
- Explicit Nudity (explicit_nudity)
- Suggestive (suggestive)
- Violence (violence)
- Visually Disturbing (visually_disturbing)
- Rude Gestures (rude_gestures)
- Drugs (drugs)
- Tobacco (tobacco)
- Alcohol (alcohol)
- Gambling (gambling)
- Hate Symbols (hate_symbols)
The default moderation confidence level to reject an image is 0.5 for all categories, unless specifically overridden. All images classified by Amazon Rekognition with a value greater than the moderation confidence level (in any of the categories) are classified as 'rejected', otherwise their status is set to 'approved'.
Cost
| Add-on | Phase | Plan | Monthly * | Monthly Price |
|---|---|---|---|---|
| Rekognition Auto Tagging | 1 | Bronze | 7.5k | $10 |
| Rekognition AI Moderation | 2 | Bronze | 7.5k | $10 |
Yearly cost of $240 with both add-ons enabled.