Award motivation:

“it is advantageous for anyone involved in the process of searching and using archive material and is open to new developments in the future. The jury strongly appreciates that the project brings the great potential of an artificial intelligence technology to a practically applicable level”

Visual search is a search modality in which the user proposes a visual example (e.g., a photograph or a video frame) to the search system as a reference for the search in place of textual expressions. The result of the search is therefore the set of contents that visually correspond to the query. The state of the art of these technologies is now at a very advanced level, such as to make this process extremely effective and efficient. The use of these technologies in the world of multimedia archives, and more, is therefore an opportunity to discover and make the most of the contents, further optimizing the costs of annotation and documentation. For many years CRITS has been following the development of these technologies, experimenting with both standard solutions (ISO / IEC 15938 – Part 13 “Compact Descriptors for Visual Search”), and those provided by the market. These studies have made it possible to effectively select the most promising technical solutions and to start a fruitful collaboration with RAI Teche for a pilot case applied to the company’s archives. The solution developed by RAI Teche, winner of the FIAT / IFTA award, is based on a cutting-edge commercial visual search engine, now integrated by Teche in the company’s media management processes.

“Visual Search: Requirements, Architectures and Use Cases to Visually Search through Broadcast Programmes“, FIAT/IFTA World Conference, 2018.

“Experimental system for visual search on broadcast archives“, CRITS Project.

Federico Maria Pandolfi, Davide Desirello, Luca Martini, Rosario Di Stabile: A Framework for Visual Search in Broadcast archives. IIR 2017: 28-39.