“A picture is worth a thousand words”, this is one of the main ideas behind content based image retrieval systems. We all use text based search engines like Google but think about how searching based on an image query can massively simplify the process of retrieving information about a particular product, the way we lookup information in the browser, or the way we organise and search in our personal collection of images or videos. Eva’s research consists of  using deep learning models to analyse the image content and create efficient and scalable visual search engines. She focuses in the need of developing systems that are generic, so one can look for a particular building but also for a particular logo or person. For that she investigates the usage of human-based attention mechanisms (saliency models) to detect the most relevant parts on the image and enhance the final image representation to make the search more effective.