Bayesian nonparametric approaches to abnormality detection in video surveillance

In data science, anomaly detection is the process of identifying the items, events or observations which do not conform to expected patterns in a dataset. As widely acknowledged in the computer vision community and security management, discovering suspicious events is the key issue for abnormal detection in video surveil-lance. The important steps in identifying such events include stream data segmentation and hidden patterns discovery. However, the crucial challenge in stream data segmenta-tion and hidden patterns discovery are the number of coherent segments in surveillance stream and the number of traffic patterns are unknown and hard to specify. Therefore, in this paper we revisit the abnormality detection problem through the lens of Bayesian nonparametric (BNP) and develop a novel usage of BNP methods for this problem. In particular, we employ the Infinite Hidden Markov Model and Bayesian Nonparamet-ric Factor Analysis for stream data segmentation and pattern discovery. In addition, we introduce an interactive system allowing users to inspect and browse suspicious events.

Piloting the feasibility of head-mounted video technology to augment student feedback during simulated clinical decision-making: an observational design pilot study

BACKGROUND: Clinical decision-making is a complex activity that is critical to patient safety. Simulation, augmented by feedback, affords learners the opportunity to learn critical clinical decision-making skills. More detailed feedback following simulation exercises has the potential to further enhance student learning, particularly in relation to developing improved clinical decision-making skills. OBJECTIVE: To investigate the feasibility of head-mounted video camera recordings, to augment feedback, following acute patient deterioration simulations. DESIGN: Pilot study using an observational design. METHODS: Ten final-year nursing students participated in three simulation exercises, each focussed on detection and management of patient deterioration. Two observers collected behavioural data using an adapted version of Gaba's Clinical Simulation Tool, to provide verbal feedback to each participant, following each simulation exercise. Participants wore a head-mounted video camera during the second simulation exercise only. Video recordings were replayed to participants to augment feedback, following the second simulation exercise. Data were collected on: participant performance (observed and perceived); participant perceptions of feedback methods; and head-mounted video camera recording feasibility and capability for detailed audio-visual feedback. RESULTS: Management of patient deterioration improved for six participants (60%). Increased perceptions of confidence (70%) and competence (80%), were reported by the majority of participants. Few participants (20%) agreed that the video recording specifically enhanced their learning. The visual field of the head-mounted video camera was not always synchronised with the participant's field of vision, thus affecting the usefulness of some recordings. CONCLUSION: The usefulness of the video recordings, to enhance verbal feedback to participants on detection and management of simulated patient deterioration, was inconclusive. Modification of the video camera glasses, to improve visual-field synchronisation with participants' actual visual field, is recommended to further explore this technology for enhancing student performance.

Demonstration of Projection

A person is shown demonstrating projection of an image at the Lithographic Technical Forum. Black and white photograph.

A inovação do AMV: o Anime Music Video e suas relações com a pós-modernidade e a cibercultura

Dissertação apresentada ao Programa de Mestrado em Comunicação da Universidade Municipal de São Caetano do Sul - USCS

Captura e visualização de video 3d em tempo real

Vídeos são dos principais meios de difusão de conhecimento, informação e entretenimento existentes. Todavia, apesar da boa qualidade e da boa aceitação do público, os vídeos atuais ainda restringem o espectador a um único ponto de vista. Atualmente, alguns estudos estão sendo desenvolvidos visando oferecer ao espectador maior liberdade para decidir de onde ele gostaria de assistir a cena. O tipo de vídeo a ser produzido por essas iniciativas tem sido chamado genericamente de vídeo 3D. Esse trabalho propõe uma arquitetura para captura e exibição de vídeos 3D em tempo real utilizando as informações de cor e profundidade da cena, capturadas para cada pixel de cada quadro do vídeo. A informação de profundidade pode ser obtida utilizando-se câmeras 3D, algoritmos de extração de disparidade a partir de estéreo, ou com auxílio de luz estruturada. A partir da informação de profundidade é possível calcular novos pontos de vista da cena utilizando um algoritmo de warping 3D. Devido a não disponibilidade de câmeras 3D durante a realização deste trabalho, a arquitetura proposta foi validada utilizando um ambiente sintético construído usando técnicas de computação gráfica. Este protótipo também foi utilizado para analisar diversos algoritmos de visão computacional que utilizam imagens estereoscópias para a extração da profundidade de cenas em tempo real. O uso de um ambiente controlado permitiu uma análise bastante criteriosa da qualidade dos mapas de profundidade produzidos por estes algoritmos, nos levando a concluir que eles ainda não são apropriados para uso de aplicações que necessitem da captura de vídeo 3D em tempo real.

Inexpensive solution for real-time video and image stitching

Image stitching is the process of joining several images to obtain a bigger view of a scene. It is used, for example, in tourism to transmit to the viewer the sensation of being in another place. I am presenting an inexpensive solution for automatic real time video and image stitching with two web cameras as the video/image sources. The proposed solution relies on the usage of several markers in the scene as reference points for the stitching algorithm. The implemented algorithm is divided in four main steps, the marker detection, camera pose determination (in reference to the markers), video/image size and 3d transformation, and image translation. Wii remote controllers are used to support several steps in the process. The built‐in IR camera provides clean marker detection, which facilitates the camera pose determination. The only restriction in the algorithm is that markers have to be in the field of view when capturing the scene. Several tests where made to evaluate the final algorithm. The algorithm is able to perform video stitching with a frame rate between 8 and 13 fps. The joining of the two videos/images is good with minor misalignments in objects at the same depth of the marker,misalignments in the background and foreground are bigger. The capture process is simple enough so anyone can perform a stitching with a very short explanation. Although real‐time video stitching can be achieved by this affordable approach, there are few shortcomings in current version. For example, contrast inconsistency along the stitching line could be reduced by applying a color correction algorithm to every source videos. In addition, the misalignments in stitched images due to camera lens distortion could be eased by optical correction algorithm. The work was developed in Apple’s Quartz Composer, a visual programming environment. A library of extended functions was developed using Xcode tools also from Apple.

A low-cost 3D reconstruction system using a single-shot projection of a pattern matrix

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

IFTrace: Video segmentation of deformable objects using the Image Foresting Transform

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

LEARNING A COMPLEX MOTOR SKILL FROM VIDEO and POINT-LIGHT DEMONSTRATIONS

The aim of this Study was to compare the learning process of a highly complex ballet skill following demonstrations of point light and video models 16 participants divided into point light and video groups (ns = 8) performed 160 trials of a pirouette equally distributed in blocks of 20 trials alternating periods of demonstration and practice with a retention test a day later Measures of head and trunk oscillation coordination d1 parity from the model and movement time difference showed similarities between video and point light groups ballet experts evaluations indicated superiority of performance in the video over the point light group Results are discussed in terms of the task requirements of dissociation between head and trunk rotations focusing on the hypothesis of sufficiency and higher relevance of information contained in biological motion models applied to learning of complex motor skills

Positive leader characteristics from high-speed video observations

Based on analyses of high-speed video recordings of cloud-to-ground lightning in Brazil and the USA, the characteristics of positive cloud-to-ground (+CG) leaders are presented. The high frame rates permitted the average, 2-dimensional speeds of development along the paths of the channels to be resolved with good accuracy. The values range from 0.3 to 6.0 x 10(5) ms(-1) with a mean of 2.7 x 10(5) ms(-1). Contrary to what is usually assumed, downward +CG leader speeds are similar to downward -CG leader speeds. Our observations also show that the speeds tend to increase by a factor of 1.1 to 6.5 as they approach the ground. The presence of short duration, recoil leaders (RLs) during the development of positive leaders reveal a highly branched structure that is not usually recorded when using conventional photographic and video cameras. The existence of the RLs may help to explain observations of UHF-VHF radiation during the development of +CG flashes.

New class of spin projection operators for 3D models

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

PHENOMENOLOGICAL APPLICATION OF THE PROJECTION-OPERATOR METHOD AND ITS CONNECTION WITH CRITICAL-DYNAMICS

We consider the dynamics of a system of interacting spins described by the Ginzburg-Landau Hamiltonian. The method used is Zwanzig's version of the projection-operator method, in contrast to previous derivations in which we used Mori's version of this method. It is proved that both methods produce the same answer for the Green's function. We also make contact between the projection-operator method and critical dynamics.