Last updated: Sep-26-2023
The insights gained from running AI analysis on your assets serve various asset management purposes, such as better decision-making and enhance asset searchability. You can request diverse AI image analysis types to categorize, label, and tag your assets.
Furthermore, you can add insights by utilizing the related assets feature. This allows you to manually establish connections between associated assets, making it easier for other users to discover and locate related assets.
You can view the results of automated AI analyses performed on an asset as well as perform new ones. Cloudinary runs some of these analyses automatically on every asset. Others require registration to an add-on. If you are not already registered for a particular add-on, you can register for the add-on's free tier and run the engine on the displayed asset directly from this pane.
To view and perform AI analysis on an asset, open the asset management drill-down page by either double-clicking the asset or selecting Open from the asset (3-dot) options menu and select the Analysis tab.
All analysis data is stored with your asset and can be used as the basis for a variety of asset management features:
Auto-tagging - Utilizes one or more auto-tagging add-ons, each of which apply deep learning AI models to identify types of objects in your image and return the list of categories it finds. You can adjust the confidence level to view additional categories or limit the list to those with a higher confidence level. You can then mark specific categories to add them as tags on your asset, or choose the option to add all the displayed categories as tags.
Face and Celebrity detection - Utilizes various algorithms and AI engines to determine the number and location of faces in your image. For celebrity detection, it also indicates the names of any celebrities detected and the accuracy confidence level. Each face/celebrity detection option displays the coordinates of the faces it finds in the form of color-coded frames within the preview area on the right. You can use the show/hide buttons for each engine to show or hide these frames. The detected faces data can be used for a number of face-detection options. For example, this data can be used to add overlays to every captured face, to blur or pixelate faces, or to apply face-detection based cropping so that you never crop out a face, and more. For the built-in face detection, you also have the option to adjust, add, or remove face coordinates (frames) and save the new face data with your image.
Area of interest - Enables you to define a custom-area of interest. This area will then be taken into account, or in some cases override other special areas, when Cloudinary programmatically crops, applies overlays, or performs other area-based transformations.
OCR - Captures text inside your images. Each phrase (line of text) that it captures is displayed in the tab and highlighted in the preview area. Similar to face-detection, you can add overlays to cover text, blur or pixelate text, or ensure that text doesn't get cut off if your images are automatically cropped for delivery.
Color analysis - Gives you information about the main colors in your image. Cloudinary can use these colors to automatically add background padding to images in a color that looks good with your image. If you have access to the Image Analysis filters (premium) of the Media Library's advanced search, you can search for images based on their predominant colors. And of course your developers can retrieve these colors to implement color-related functionality in your application.
Cloudinary's related assets functionality allows users to establish relationships between assets based on specific organizational needs and criteria. By manually configuring these associations, users can create meaningful connections between assets within their library, making it easier for other users to discover and locate similar or associated assets.
Here are some use cases where you might need to use related assets:
- While looking at a logo, you want to easily access the 'Brand Guidelines' PDF, so you'll know how much spacing to use.
- You want to see a video's peripheral assets (transcript, poster, etc.)
- You want to access other images of the same product.
You can select one or more assets and relate them to another asset. The asset(s) you selected initially now have a two-way relationship with the asset you related them to. In other words, any of the assets that you selected initially will give easy access to the asset that you related them to, and the asset you related to will give easy access to all the assets you selected initially. The assets you selected initially aren't related to each other.
Here's a diagram that illustrates this relationship:
To relate assets:
- Select one or more assets in your Media Library.
- Select Relate To... from the (3-dots) option menu:
- On the asset, if you've selected one asset.
- On the asset toolbar, if you've selected more than one asset.
- Find the asset that you want to relate the selected assets to by typing the asset’s public ID in the dialog box.
- Create the relationship.
An asset's related assets are displayed in the Related... tab of the Preview pane and Manage page. From there, you can Open or Download a related asset, or you can Remove relationship so that the assets will no longer be related.