"Smart Retrieval"- the advanced multimedia search tool that makes it smart and easy to find the perfect image!
This web application uses state-of-the-art technology to provide advanced search functions for multimedia archives, including content-based image retrieval (CBIR) functions. With Smart Retrieval users can search for images in two ways: using a text description or a combination of reference image and associated text description.
What makes Smart Retrieval apart is its state-of-the-art computer vision component, which uses a neural network to associate the text describing the desired image content with the image pixels. This allows us to perform conditional image retrieval to search for an image using an image example and an additional textual description that describes a change from the reference image content.
Smart Retrieval can also be used to perform tagging, or annotation of images, using a zero-shot learning approach. This means that even small museums or institutions with limited resources can benefit from our technology without having to collect large amounts of training data.
Discover and test
Try a demo of Smart Retrieval at the following link:
http://solaris.micc.unifi.it
