|Research Area:||Robotics and Perception|
|Phone or fax:||+390755853682|
Elisa Ricci received her M.S. and Ph.D. degree in Electrical Engineering from the University of Perugia respectively in 2004 and 2008. Her M.S. thesis with title "On the Use of Neural Networks to Solve the Reverse Modeling Problem for Scanning Capacitance Microscopy" was developed during a period of six month internship at the Integrated Systems Laboratory (IIS) of the Swiss Federal Institute of Technology (ETH) in Zurich. During her PhD she spent one year as a visiting student at University of Bristol working in the Pattern Analysis and Intelligent Systems group. Her PhD dissertation was about a novel theoretical and algorithmic framework for learning with structured data. After her PhD she joined the Idiap Research Institute, Martigny, Switzerland, where she stayed one year as a postdoctoral researcher. Then she moved at Fondazione Bruno Kessler, where she worked two years as junior researcher in the TEV team. From January 2011 she is an assistant professor at Dipartimento di Ingegneria Elettronica e dell'Informazione at Università di Perugia. Her research interests are mainly in the areas of computer vision, machine learning and robotics.
Università degli Studi di Perugia
Facoltà di Ingegneria
Dipartimento di Ingegneria Elettronica e dell'Informazione
Via Duranti 93
I-06125 Perugia Italy
Phone:+39 075 585 3682
Fax:+39 075 585 3654
Personal Webpage: https://sites.google.com/site/elisaricciunipg/home
We study novel approaches for automatically discovering spatio-temporal patterns in complex dynamic scenes. Similarly to recent non-object centric methods, we use low level visual cues to detect atomic activities and then construct clip histograms. Differently from previous works, we formulate the task of discovering high level activity patterns as a prototype learning problem where the correlation among atomic activities is explicitly taken into account when grouping clip histograms. At the core of our approaches there is a convex optimization problem which allows us to efficiently extract patterns at multiple levels of detail. The project is in collaboration with DISI, University of Trento.
Visual Loop Closure Detection
We study learning approaches for improving current place recognition systems based on the bag-of-words paradigm. The place recognition module is a fundamental component in SLAM systems, as incorrect loop closures may result in severe errors in trajectory estimation. In the case of appearance-based methods the bag-of-words approach is typically employed for recognizing locations. In this work we study novel methods for improving loop closures detection performance by adopting a set of visual words weights, learned offline accordingly to a discriminative criterion. Our experiments, conducted on publicly available datasets, demonstrate that the discriminative weights lead to loop closures detection results that are more accurate than the traditional bag-of-words method and that our place recognition approach is competitive with state-of-the-art methods.