Publications by Roberto Vezzani

This paper describe the ViSOR (Video Surveillance Online Repository) repository, designed with the aim of establishing an open platform for collecting, annotating, retrieving, and sharing surveillance videos, as well as evaluating the performance of automatic surveillance systems. The repository is free and researchers can collaborate sharing their own videos or datasets. Most of the included videos are annotated. Annotations are based on a reference ontology which has been defined integrating hundreds of concepts, some of them coming from the LSCOM and MediaMill ontologies. A new annotation classification schema is also provided, which is aimed at identifying the spatial, temporal and domain detail level used. The web interface allows video browsing, querying by annotated concepts or by keywords, compressed video previewing, media downloading and uploading. Finally, ViSOR includes a performance evaluation desk which can be used to compare different annotations.

The research on innovative and natural interfaces aims at developing devices able to capture and understand the human behavior without the need of a direct interaction. In this paper we propose and describe a framework based on a sensing floor device. The pressure field generated by people or objects standing on the floor is captured and analyzed. Local and global features are computed by a low level processing unit and sent to high level interfaces. The framework can be used in different applications, such as entertainment, education or surveillance. A detailed description of the sensing element and the processing architectures is provided, together with some sample applications developed to test the device capabilities.

People re-identification is a challenging problem in surveillance and forensics and it aims at associating multiple instances of the same person which have been acquired from different points of view and after a temporal gap. Image-based appearance features are usually adopted but, in addition to their intrinsically low discriminability, they are subject to perspective and view-point issues. We propose to completely change the approach by mapping local descriptors extracted from RGB-D sensors on a 3D body model for creating a view-independent signature. An original bone-wise color descriptor is generated and reduced with PCA to compute the person signature. The virtual bone set used to map appearance features is learned using a recursive splitting approach. Finally, people matching for re-identification is performed using the Relaxed Pairwise Metric Learning, which simultaneously provides feature reduction and weighting. Experiments on a specific dataset created with the Microsoft Kinect sensor and the OpenNi libraries prove the advantages of the proposed technique with respect to state of the art methods based on 2D or non-articulated 3D body models.

The field of surveillance and forensics research is currently shifting focus and is now showing an ever increasing interest in the task of people reidentification. This is the task of assigning the same identifier to all instances of a particular individual captured in a series of images or videos, even after the occurrence of significant gaps over time or space. People reidentification can be a useful tool for people analysis in security as a data association method for long-term tracking in surveillance. However, current identification techniques being utilized present many difficulties and shortcomings. For instance, they rely solely on the exploitation of visual cues such as color, texture, and the object's shape. Despite the many advances in this field, reidentification is still an open problem. This survey aims to tackle all the issues and challenging aspects of people reidentification while simultaneously describing the previously proposed solutions for the encountered problems. This begins with the first attempts of holistic descriptors and progresses to the more recently adopted 2D and 3D model-based approaches. The survey also includes an exhaustive treatise of all the aspects of people reidentification, including available datasets, evaluation metrics, and benchmarking.

Nowadays, intelligent video surveillance has become an essential tool of the greatest importance for several security-related applications. With the growth of installed cameras and the increasing complexity of required algorithms, in-house self-contained video surveillance systems become a chimera for most institutions and (small) companies. The paradigm of Video Surveillance as a Service (VSaaS) helps distributing not only storage space in the cloud (necessary for handling large amounts of video data), but also infrastructures and computational power. This chapter will briefly introduce the motivations and the main characteristics of a VSaaS system, providing a case study where research-lab computer vision algorithms are integrated in a VSaaS platform. The lessons learnt and some future directions on this topic will be also highlighted.

The recognition of people orientation in single images is still an open issue in several real cases, when the image resolution is poor, body parts cannot be distinguished and localized or motion cannot be exploited. However, the estimation of a person orientation, even an approximated one, could be very useful to improve people tracking and re-identification systems, or to provide a coarse alignment of body models on the input images. In these situations, holistic features seem to be more effective and faster than model based 3D reconstructions. In this paper we propose to describe the people appearance with multi-level HoG feature sets and to classify their orientation using an array of Extremely Randomized Trees classifiers trained on quantized directions. The outputs of the classifiers are then integrated into a global continuous probability density function using a Mixture of Approximated Wrapped Gaussian distributions. Experiments on the TUD Multiview Pedestrians, the Sarc3D, and the 3DPeS datasets confirm the efficacy of the method and the improvement with respect to state of the art approaches.

Safety and security reasons are pushing the growth of surveillance systems, for both prevention and forensic tasks. Unfortunately, most of the installed systems have recording capability only, with quality so poor that makes them completely unhelpful. This chapter will introduce the concepts of modern systems for Intelligent Video Surveillance (IVS), with the claim of providing neither a complete treatment nor a technical description of this topic but of representing a simple and concise panorama of the motivations, components, and trends of these systems. Different from CCTV systems, IVS should be able, for instance, to monitor people in public areas and smart homes, to control urban traffi c, and to identity assessment for security and safety of critical infrastructure.

In this paper we introduce a novel surveillance system, which uses 3D information extracted from multiple cameras to detect, track and re-identify people. The detection method is based on a 3D Marked Point Process model using two pixel-level features extracted from multi-plane projections of binary foreground masks, and uses a stochastic optimization framework to estimate the position and the height of each person. We apply a rule based Kalman-filter tracking on the detection results to find the object-to-object correspondence between consecutive time steps. Finally, a 3D body model based long-term tracking module connects broken tracks and is also used to re-identify people

AD-HOC (Appearance Driven Human tracking with Occlusion Classification) is a complete framework for multiple people tracking in video surveillance applications in presence of large occlusions. The appearance-based approach allows the estimation of the pixel-wise shape of each tracked person even during the occlusion. This peculiarity can be very useful for higher level processes, such as action recognition or event detection. A first step predicts the position of all the objects in the new frame while a MAP framework provides a solution for best placement. A second step associates each candidate foreground pixel to an object according to mutual object position and color similarity. A novel definition of non-visible regions accounts for the parts of the objects that are not detected in the current frame, classifying them as dynamic, scene or apparent occlusions. Results on surveillance videos are reported, using in-house produced videos and the PETS2006 test set.