Giving Computers the Gift of Sight
True artificial intelligence and reasoning still remain a scientific challenge but significant progress is being made in a number of fields that contribute to achieving this objective. Two such domains are addressed in this theme – computer vision and text mining. New techniques and algorithms in computer vision mean that computers are now able to not only recognise what they ‘see’ in videos and photos but they are able to compare what they see in detail and can even go so far as to select what images are aesthetically ‘better’ than others. Xerox state-of-the-art technology in fine grained image recognition and classification is currently being applied to solve problems in the field of healthcare, transportation, retail and document workflow imaging. International benchmarking results confirm that Xerox is a recognised leader in the field without being computationally intensive requiring racks of processors thus making its approach sustainable, cost effective and flexible.
Teaching computers to read and understand text like humans means we need to be able to program them to not only decode characters, words and sentences but to do so within a context which allows them to make connections between and across the words, facts and situations described by the language. This ultimately means being able to understand the semantic content or meaning. The text mining research team in Europe has developed technology that can today identify relations between texts in large collections of unstructured documents and also associate these with a meaning that can distinguish it from other meanings. By combining software with specialised terminology and domain specific rules e.g. medical, financial, pharmaceutical it is capable of uncovering critical information and can even detect drug side effects, illnesses and diseases.
- Computer Vision introduction, Florent Perronnin
- Large-scale image retrieval, Florent Perronnin
- Visual aesthetics, Luca Marchesotti
- Video Monitoring of Patient Vital Signs, Raja Bala
- Video Compression for Efficient Search and Retrieval, Edgar A. Bernal
- Understanding Text and Unstructured Information and detection of hospital acquired infections, Caroline Hagège, Denys Proux