Textual & Visual Pattern Analysis
The Textual and Visual Pattern Analysis area seeks to create technology that makes everyday interaction with visual and textual content simple and effective. Our research is the result of combining skills in machine learning, pattern recognition, text mining and computer vision. This research is often pursued in collaboration with external partners from industry, government and academia.
The main research challenges the team is currently addressing are:
- Hybrid Information Access: Dealing with the large-scale, temporally varying, heterogeneous content collections. The hybrid nature of information is not limited to multiple modes such as images and text but also to the increasingly frequent case where the metadata is the data, as in the case of user tags or comments.
- Applied Visual Aesthetics: Aesthetic analysis goes beyond traditional image quality analysis in that it also needs to take into account subjective, social, emotional and semantic dimensions. The recent advances in semantic analysis, social networking services and machine learning enable the development of aesthetic models.
Selected Projects
- Image Categorization and Retrieval
- Hybrid Information Retrieval
- Handwritten Wordspotting
- Aesthetic Quality Assessment
Government Projects and other Open Innovation Initiatives
- PinView: An information navigation system for visual data that exploits implicit feedback from its users
- Shaman: long-term archiving
- OMNIA:Analysis, classification and organization of multilingual rich-content documents
- Fragrance: text analysis, social network analysis
- SYNC3: text clustering, sentiment analysis, opinion mining
Past Projects
- Automatic Image Enhancement
- Mobile Document Imaging
- Text Categorization and Clustering
- SMART
- InfoM@gic
Publications
People