Eduardo Monari

A knowledge-based camera selection approach for object tracking in large sensor networks

In this paper an approach for dynamic sensor selection in large video-based sensor networks for the purpose of multi-camera object tracking is presented. The sensor selection approach is based on computational geometry algorithms and is able to determine task-relevant cameras (camera cluster) by evaluation of geometrical attributes, given the last observed object position, the sensor configurations and the environment model. Hereby, a special goal of this algorithm is to determine the minimum number of sensors needed to relocate an object, even if the object is temporarily out of sight (e.g., by non-overlapping sensor coverage). It will be shown that the algorithm enables self-organizing tracking approaches to perform optimal camera selection in a highly efficient way. In particular, the approach is applicable to very large camera networks and leads to a highly reduced network and processor load for multi-camera tracking.

A real-time image-to-panorama registration approach for background subtraction using pan-tilt-cameras

While for static cameras several background subtraction approaches have been developed in the past, for non-static pan/tilt cameras efficient and robust motion detection is still a challenging task. Known approaches use image-to-image registration methods to generate a panorama background model of the scene, which spans a joint pixel coordinate system for later background estimation and subtraction. However, for a real-time panorama-based background subtraction a highly efficient image-to-panorama registration is needed. For this purpose, in this paper a key-frame representation of the panorama image is proposed and a strategy for fast global homography estimation in large panorama images is presented.

Fast, robust and automatic 3D face model reconstruction from videos

This paper presents a fully automatic system that recovers 3D face models from sequences of facial images. Unlike most 3D Morphable Model (3DMM) fitting algorithms that simultaneously reconstruct the shape and texture from a single input image, our approach builds on a more efficient least squares method to directly estimate the 3D shape from sparse 2D landmarks, which are localized by face alignment algorithms. The inconsistency between self-occluded 2D and 3D feature positions caused by head pose is ad-dressed. A novel framework to enhance robustness across multiple frames selected based on their 2D landmarks combined with individual self-occlusion handling is proposed. Evaluation on groundtruth 3D scans shows superior shape and pose estimation over previous work. The whole system is also evaluated on an “in the wild” video dataset [12] and delivers personalized and realistic 3D face shape and texture models under less constrained conditions, which only takes seconds to process each video clip.

Towards robust cascaded regression for face alignment in the wild

Most state-of-the-art solutions for localizing facial feature landmarks build on the recent success of the cascaded regression framework [7, 15, 34], which progressively predicts the shape update based on the previous shape estimate and its feature calculation. We revisit several core aspects of this framework and show that proper selection of regression method, local image feature and fine-tuning of further fitting strategies can achieve top performance for face alignment using the generic cascaded regression algorithm. In particular, our strongest model features Iteratively Reweighted Least Squares (IRLS) [18] for training robust regressors in the presence of outliers in the training data, RootSIFT [2] as the image patch descriptor that replaces the original Euclidean distance in SIFT [24] with the Hellinger distance, as well as coarse-to-fine fitting and in-plane pose normalization during shape update. We show the benefit of each proposed improvement by extensive individual experiments compared to the baseline approach [34] on the LFPW dataset [4]. On the currently most challenging 300-W dataset [28] and COFW dataset [4], we report state-of-the-art results that are superior to or on par with recently published algorithms.

A soft-biometrics dataset for person tracking and re-identification

In this work we present a new dataset for the tasks person detection, tracking, re-identification, and soft-biometric attribute detection in surveillance data. The dataset was recorded over three days and consists of more than 30 individuals moving through a network of seven cameras. Person tracks are labeled with consistent IDs as well as soft-biometric attributes, such as a description of the clothing, gender, or height. Persons in the video data alter their appearance by changing clothes or wearing accessories. A second, clothing specific ID of each track allows for the evaluation of re-identification with or without the presence of clothing changes. In addition to video and camera calibration data, we provide evaluation protocols, tools and baseline results for each of the four tasks.

Color Constancy Using Shadow-Based Illumination Maps for Appearance-Based Person Re-identification

Robust people re-identification is one of the most challenging task, and still an unsolved problem for several applications in video surveillance. A large number of approaches use colors as main features for object description, which are in fact important cues for re-identification. However, colors captured by a camera suffer from unknown and changing global and local illumination conditions in the scene. Thus, color constancy is an essential pre-condition for robust color-based person re-identification. In this paper we introduce a new approach for automated estimation and compensation of local illumination in the scene. The proposed approach allows for handling of multiple light sources in the scene, and to compensate backlight illumination simultaneously. Both are continuous problems with high relevance to practical use of color-based approaches in video surveillance.

Task-Oriented Object Tracking in Large Distributed Camera Networks

In this paper a task-oriented approach for object trackingin large distributed camera networks is presented. Thiswork includes three main contributions. First a generic processframework is presented, which has been designed fortask-oriented video processing. Second, system componentsof the task-oriented framework needed for the task of multicameraperson tracking are introduced in detail. Third, foran efficient task-oriented processing in large camera networksthe capability of dynamic sensor scheduling by themulti-camera tracking processes is indispensable. For thispurpose an efficient sensor selection approach is proposed.

Adaptive Contour Fitting for Pose-Invariant 3D Face Shape Reconstruction.

Motivation Direct reconstruction of 3D face shape—solely based on a sparse set of 2D feature points localized by a facial landmark detector— offers an automatic, efficient and illumination-invariant alternative to the widely known analysis-by-synthesis framework, which is extremely timeconsuming considering the enormous parameter space for both shape and photometric properties. Given 2D landmarks y and their correspondence on the 3D Morphable Model (3DMM), the 3D shape can be recovered by minimizing the distance between 2D and the projected 3D landmarks

Capturing ground truth super-resolution data

Super-resolution (SR) offers an effective approach to boost quality and details of low-resolution (LR) images to obtain high-resolution (HR) images. Despite the theoretical and technical advances in the past decades, it still lacks plausible methodology to evaluate and compare different SR algorithms. The main cause to this problem lies in the missing ground truth data for SR. Unlike in many other computer vision tasks, where existing image datasets can be utilized directly, or with a little extra annotation work, evaluating SR requires that the dataset contain both LR and the corresponding HR ground truth images of the same scene captured at the same time. This work presents a novel prototype camera system to address the aforementioned difficulties of acquiring ground truth SR data. Two identical camera sensors equipped with a wide-angle lens and a telephoto lens respectively, share the same optical axis by placing a beam splitter in the optical path. The back-end program can then trigger their shutters simultaneously and precisely register the region of interests (ROIs) of the LR and HR image pairs in an automated manner free of sub-pixel interpolation. Evaluation results demonstrate the special characteristics of the captured ground truth HR-LR face images compared to the simulated ones. The dataset is made freely available for noncommercial research purposes.

Dynamic Sensor Selection for Single Target Tracking in Large Video Surveillance Networks

In this paper an approach for dynamic camera selectionin large video-based sensor networks for the purposeof multi-camera object tracking is presented. The sensor selectionapproach is based on computational geometry algorithmsand is able to determine task-relevant cameras (cameracluster) by evaluation of geometrical attributes, giventhe last observed object position, the sensor configurationsand a building map. A special goal of this algorithm is theefficient determination of the minimum number of sensorsneeded to relocate an object, even if the object is temporarilyout of sight. In particular, the approach is applicablein camera networks with overlapping and non-overlappingfield of views as well as with static and non-static sensors.