Developed initially to detect selected objects in seized images, f2 Object has been designed to find images of the same object, even if it’s been resaved, copied, or cropped, in any database containing images. This is regardless of:

• Image format
• Cropping
• Resizing
• Position in the image
• Size in the image (subject to sufficient resolution)

Example applications of f2 Object include vehicles, logos, flags, buildings, etc, as well as objects added temporarily which are specific to a particular search.

Key features include:

• f2 Object is designed to find and encode common patterns and characteristics across images, including in different formats
• Matches any size of object from very large to very small
• Partially concealed objects can still be matched
• Matches are displayed as thumbnails together with location and Match%
• Image enhancement maximizes matching performance with small and lo-res images
• Probe and database images can match at up to a 35° rotational difference
• Straightforward to connect to existing databases containing images, in read-only mode
• Side-by-side comparison of “probe” and matching images, including magnification, for visual inspection
• Clicking on any matching image will lead back to the original database record
• Automatic enrollment of the characteristics of all images in a database, and ongoing updates to it

There’s no limit to the size of database that can be searched. f2 Object will automatically analyze and enroll images in an existing database, as well as import and enroll images held in folders. It will automatically detect and enroll new images that are added to external databases.

The object matching capability is available as a stand-alone/networked application, as a .NET SDK, and as a web service. It can be configured to search databases across multiple machines on a VPN, under appropriate controls.

Object matching is part of the f2 image recognition suite, which also includes full and partial face recognition, corpse identification, tattoo recognition, and scene recognition. All modules can link together. There’s no limit to the size of databases that can be enrolled and searched. Objects in folders can be automatically imported, enrolled, and searched.