Publications by Rita Cucchiara

The common paradigm employed for object detection is the sliding window (SW) search. This approach generates grid-distributed patches, at all possible positions and sizes, which are evaluated by a binary classifier: the trade-off between computational burden and detection accuracy is the real critical point of sliding windows; several methods have been proposed to speed up the search such as adding complementary features. We propose a paradigm that differs from any previous approach, since it casts object detection into a statistical-based search using a Monte Carlo sampling for estimating the likelihood density function with Gaussian kernels. The estimation relies on a multi-stage strategy where the proposal distribution is progressively refined by taking into account the feedback of the classifiers. The method can be easily plugged in a Bayesian-recursive framework to exploit the temporal coherency of the target objects in videos. Several tests on pedestrian and face detection, both on images and videos, with different types of classifiers (cascade of boosted classifiers, soft cascades and SVM) and features (covariance matrices, Haar-like features, integral channel features and histogram of oriented gradients) demonstrate that the proposed method provides higher detection rates and accuracy as well as a lower computational burden w.r.t. sliding window detection.

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.

This paper presents an innovative approach for detecting and localizing duplicate objects in pick-and-place applications under extreme conditions of occlusion, where standard appearance-based approaches are likely to be ineffective. The approach exploits SIFT keypoint extraction and mean shift clustering to partition the correspondences between the object model and the image onto different potential object instances with real-time performance. Then, the hypotheses of the object shape are validated by a projection with a fast Euclidean transform of some delimiting points onto the current image. Moreover, in order to improve the detection in the case of reflective or transparent objects, multiple object models (of both the same and different faces of the object) are used and fused together. Many measures of efficacy and efficiency are provided on random disposals of heavily-occluded objects, with a specific focus on real-time processing. Experimental results on different and challenging kinds of objects are reported. © 2012 Elsevier B.V. All rights reserved.

Image collections are searched in common retrieval systems in many different ways, but the typical presentation is by means of a grid styled view. In this paper we try to suggest a novel use of relevance feedback as a tool to warp the view and allow the user to spatially navigate the image collection, and at the same time focus on his retrieval aim. This is obtained by the use of a distance based space warping on the 2D projection of the distance matrix.

For quite some time, libraries, document and historical centers from opposite corners of the world have been the caretakers of our rich and assorted social legacy. They have protected and furnished access to the testimonies of knowledge, beauty and inspiration, such as sculptures, paintings, music and literature. The new information technologies have created unbelievable opportunities to make this common heritage more accessible for all. Culture is following the digital path and “memory institutions” are adapting the way in which they communicate with their public. Multimedia technologies have recently created the conditions for a true revolution in the cultural heritage area, with reference to the study, valorization, and fruition of artistic works. New multimedia technologies shall be able to be utilized to plan unique approaches to the perception and fulfillment of the masterful legacy, for instance, through smart cultural objects and new interfaces with the backing of items such as story-telling, gaming and learning.All the plurality of masterpieces (paintings, books, manuscripts, even photos of sculptures and architecture) can be effectively embedded into a unique ``paradigm'' through digitization. This allows a significant reduction in costs, an enormous expansion of public accessibility (and therefore income), and at the same time a tremendous freedom for data elaboration. In brief, digitization enhances pleasure for the public and usefulness to experts on cultural heritage assets.

Artistic image collections are usually managed via textual metadata into standard content management systems. More sophisticated searches can be performed using image retrieval technologies based on visual content. Nevertheless, the problem of the information presentation remains. In this paper we try to move beyond the classic grid-styled presentation model, suggesting a novel use of relevance feedback as a navigation tool. Relevance feedback is therefore used to warp the view and allow the user to spatially navigate the image collection, and at the same time focus on his retrieval aim. This is obtained exploiting a distance based space warping on the 2D projection of the distance matrix. Multitouch gestures are employed to provide feedbacks by natural interaction with the system.

Understanding social and collective people behaviour in open spaces is one of the frontier of modern video surveillance. Many sociological theories, and proxemics in particular, have been proved their validity as a support for classifying and interpreting human behaviour. Proxemics suggest some simple but effective behavioural rules, useful to understand what people are doing and their social involvement with other individuals. In this paper we propose to extend the proxemics analysis along the time and provide a solution for analysing sequences of proxemic states computed between trajectories of people pairs (dyads). Trajectories, computed from videosurveillance videos, are first analysed and converted to a sequence of symbols according to proxemic theory. Then an elastic measure for comparing those sequences is introduced. Finally, interactions are classified both in an off-line unsupervised way and in an on-line fashion. Results on videosurveillance data, demonstrate that sequences of proxemic states can be effective in characterizing mutual interactions and experiments in capturing the most frequent dyads interactions and on-line classifying them when a labelled training set is available are proposed.

This paper describes the design and development of a Veiling Luminance estimation system based on the use of a CMOS image sensor, fully implemented on FPGA. The system is composed of the CMOS Image sensor, FPGA, DDR SDRAM, USB controller and SPI (Serial Peripheral Interface) Flash. The FPGA is used to build a system-on-chip integrating a soft processor (Xilinx MicroBlaze) and all the hardware blocks needed to handle the external peripherals and memory. The soft processor is used to handle image acquisition and all computational tasks need to compute the Veiling Luminance value. The advantages of this single chip FPGA implementation include the reduction of the hardware requirements, power consumption, and system complexity. The problem of the high dynamic range images have been addressed with multiple acquisitions at different exposure times. Vignetting, radial distortion and angular weighting, as required by veiling luminance definition, are handled by a single integer look-up table (LUT) access. Results are compared with a state of the art certified instrument.

The interest of the research community in creating reference datasets for performance analysis is always very high. Although new datasets, collecting large amounts of video footage are spreading in surveillance and forensics, few bench-marks with annotation data are available for testing specific tasks and especially for 3D/multi-view analysis. In this paper we present 3DPeS, a new dataset for 3D/multi- view surveillance and forensic applications. This has been designed for discussing and evaluating research results in people re-identification and other related activities (people detection, people segmentation and people tracking). The new assessed version of the dataset contains hundreds of video sequences of 200 people taken from a multi-camera distributed surveillance system over several days, with different light conditions; each person is detected multiple times and from different points of view. In surveillance scenarios, the dataset can be exploited to evaluate people reacquisition, 3D body models and people activity reconstruction algorithms. In forensics it can be adopted too, by relaxing some constraints (e.g. real time) and neglecting some information (e.g. calibration). Some results on this new dataset are presented using state of the art methods for people re-identification as a benchmark for future comparisons.