The main problem solved by image classification is sorting and grouping images into classes and categories.
Image classification is an extremely useful and very powerful tool in today's world of machine vision. Surfy Vision Image Classification segments images into groups at lightning speed using a variety of models and scores the results.
Surfy Vision Image Classification uses already predefined Surfy Vision models and the Surfy Search Engine Database for image recognition and grouping. If no matches are found or the data obtained is not conclusive enough, the Surfy algorithms start searching for similar images on the global Internet.
The median search time for this scenario is currently 1.15s. If the image classification is successful, the Surfy Vision algorithms will extend the current model or create a new additional model, according to the nesting of the Surfy hierarchical object structure.
During learning, Surfy's algorithms significantly increase the search speed, which at its peak reaches 0.005s per frame processed.
Flexible hierarchical image recognition model with 240x240x3 matrix allows to achieve speed of 0.005s per frame from one core on 2.3mHz CPU eliminating the need for expensive GPU processors.
Image classification does not necessarily have to consist of a large number of groups of objects, but can be fine-tuned to distinguish one from the other, such as faults in production or searching for specific items in large volumes of visual information.
All Surfy Vision products can be combined and integrated with each other, giving you the unique features and characteristics needed to achieve your business objectives.
We can customise and extend Surfy Vision Image Classification to suit your company's needs.