If all goes well after installing, you should be able to point your browser to http://localhost:31128/. This is the admin web interface, where you can check some database statistics, configuration options and perform some basic operations on the image database:
All the following operations can be done using the XML-RPC/SOAP API. Client libraries and other notes on using this API are availableonline or at the directory "api-docs/" directory of the source package.
An isk-daemon server instance can hold several database spaces. Each database space represents a separate namespace for image ids. Image ids must be unique within a database space and similarity queries are performed only within one database space.
The first step is creating one database space. Click on the "Databases" tab on the left and then on the "Create new" tab on the top:
All IDs on isk-daemon are long integers: database space IDs, image IDs, keywords. Enter '1' and click on the 'Add' button.
All database spaces are persisted to disk on a single file. The full path for this file can be changed on the config file isk-daemon.conf. By default everything is saved when isk-daemon is shut down or every 2 minutes when images are being added. You can call via API the saveDb() operations to make sure your latest changes are written to disk or click the "Save all" button on the Databases -> Existing tab of the admin web interface.
Next, you need to add images to a database space. isk-daemon only needs access to the actual image file and its data at the indexing moment. After being indexed, image files may be removed.
Click on the "Images" tab at the left, then on "Import images" tab at the top. Notice that we're operating on the database space we've just created (id=1):
Enter a directory path on the computer where isk-daemon is installed which contains images and click on the "Import" button. You should later see a message with the total number of images indexed. Since we're using the web admin UI, these images were assigned a random sequential ID at index time. Image IDs should be unique within one database space. To have control over the which ID is assigned to which image when they are indexed (so that search results make sense later) the API must be used.
The default configuration is to persist the database to disk every 2 minutes, but you can always force a save using an option on the admin web interface (Databases tab, then "Save all" button) or using the API itself.
You can test some operations on the "Images" tab on the left. Here's an example of a visual similarity search:
isk-daemon supports the concept of keywords or tags, that can be assigned to images. Keywords are not text-based, they are known to isk-daemon simply as integers for performance reasons. Keywords can be used during visual similarity searches for restricting the search scope.