Image categorization is time consuming and predominantly done manually. The amount of user generated content in the recent years requires ways to easily review and categorize photos automatically. Our Auto Categorization technology helps you save time and money by automating the process.
Imagga's Automated Image Categorization performs a really fast categorization on any image set, by performing a combination of image processing and machine learning algorithms. Imagga's solution also provides a straightforward way to train the categorizer against your company's specific image categories.
You can use some of the predefined general image classifiers that we've trained via machine learning, but you can also train your own. To create your specific classifier we need about a thousand representative and diverse training images for each category that you want the system to learn to recognize.
The image recognition engine will learn to recognize the categories, it's trained with, by building a set of local and global visual features. These features that will later activate the right categorization when a new, still unseen, image is given for classification.
Each of the images that you fed to the system is assigned a list of confidence levels, showing how confident the categorization core is when assigning the given image to each of the categories in the list. This allows you to apply certain thresholds and eventually to assign one image to more than one category.