Current advanced vision systems aiming for activity recognition are strongly dependant on the locations of particular scenes, thereby restricting the semantic properties of the places where such events occur. Thus, a task for automatic and generic categorization of semantic regions is demanded in the field.

Automatic region labeling on intercity urban scenarios

Carles Fernàndez Tena presents a method to perform automatic region labeling on intercity urban scenarios, based on the trajectories of pedestrians and vehicles observed by public webcams. As a result, the system divides the scenario into regions like crosswalks, sidewalks, roads, or waiting zones. Prior information is first modeled using a simple feature-based approach, and additional domain knowledge helps constraining spatial coherence to the results, by means of a MAP-MRF inference process. A progressive and thorough experimental validation will be presented, discussing a series of complementary steps that help enhancing the proposed framework.

In this talk the engeineer Andrea Fusiello will tackle the problem of fitting multiple instances of a model to data corrupted by noise and outliers. A solution will be proposed based on random sampling and conceptual data representation. Each data point is represented with the characteristic function of the set of random models that fit that point.

Robust multiple structures estimation and architectural modelling

A tailored agglomerative clustering, called J-linkage, is used to group points belonging to the same model. The method does not require prior specification of the number of models, nor it necessitates parameter tuning. Finally, I will touch upon an application of this technique to unsupervised reconstruction of architectural models in terms of high-level primitives.

The ability to interpret the semantics of objects and actions, their individual geometric attributes as well as their spatial and temporal relationships within the environment is essential for an intelligent visual system and extremely valuable in numerous applications. In visual recognition, the problem of categorizing generic objects is a highly challenging one. Single objects vary in appearances and shapes under various photometric (e.g. illumination) and geometric (e.g. scale, view point, occlusion, etc.) transformations. Largely due to the difficulty of this problem, most of the current research in object categorization has focused on modeling object classes in single (or nearly single) views. But our world is fundamentally 3D and it is crucial that we design models and algorithms that can handle such appearance and pose variability.

Ing. Silvio Savarese: Visual recognition in the three dimensional world

In this talk I will introduce a novel framework for learning and recognizing 3D object categories and their poses. Our approach is to capture a compact model of an object category by linking together diagnostic parts of the objects from different viewing points. The resulting model is a summarization of both the appearance and geometry information of the object class. Unlike earlier attempts for 3D object categorization, our framework requires minimal supervision and has the ability to synthesize unseen views of an object category. Our results on categorization show superior performances to state-of-the-art algorithms on the largest dataset up to date. I will conclude the talk with final remarks on the relevance of the proposed research for a number of applications in mobile vision.