The amount of information available online has increased over the last years. In a first stage, network users were mostly consumers of information, while the data producers were a very limited number of specific nodes. Gradually this approach has changed. Currently, Internet users often adopt both roles (producer and consumer of information), increasing dramatically the volume of available content. In this context, one of the most serious problems facing the network is the search for information. For a common user, it is very important:
- Being able to easily locate the content needed.
- Being able to interact with the machine in a user-friendly manner.
This problem has traditionally been treated using techniques based on text. However, we do not have yet efficient tools for doing semantic searches for audiovisual data. In particular, we work with algorithms that analyze, process, learn and recognize information present in digital images. We are interested in:
- Algorithms for automatic object recognition and scene understanding in natural images, for its subsequent use in classification problems in natural environments.
- Algorithms for recognition of gestures, facial expressions and non verbal language using images and videos of people, in order to build user-friendly interfaces for human-machine interaction, and analyze social interactions between people that arise from social honest signals.
The Computer Vision and Artificial Intelligence knowledge fields will help us to develop methods for these purposes.