Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks123 from public data
- 5,740
- Exploiting the Circulant Structure of Tracking-by-Detection with Kernels2,365
Using the well-established theory of Circulant matrices, this work provides a link to Fourier analysis that opens up the possibility of extremely fast learning and detection with the Fast Fourier Transform, which can be done in the dual space of kernel machines as fast as with linear classifiers.
- The Visual Object Tracking VOT2014 Challenge Results588
The Visual Object Tracking challenge 2014, VOT2014, aims at comparing short-term single-object visual trackers that do not apply pre-learned models of object appearance and introduces a new dataset with full annotation of targets by rotated bounding boxes and per-frame attribute.
- Semantic Segmentation with Second-Order Pooling533
This paper introduces multiplicative second-order analogues of average and max-pooling that together with appropriate non-linearities lead to state-of-the-art performance on free-form region recognition, without any type of feature coding.
- 204
- 202
- 149
- 113
- 73
- 72
- 65
- 63
- 63
- 61
- Beyond the shortest path: Unsupervised domain adaptation by Sampling Subspaces along the Spline Flow58
- A vision-based system for measuring the displacements of large structures: Simultaneous adaptive calibration and full motion estimation53
A kinematic model-based solution to estimate simultaneously the calibration parameters of the vision system and the full-motion of large civil engineering structures, namely of long deck suspension bridges, from a sequence of stereo images captured by digital cameras is described.
- 53
- 45
- A Real-Time Driver Visual Attention Monitoring System44
A framework for analyzing video sequences of a driver and determining his level of attention is described, which relies on pupil detection to robustly track the driver's head pose and monitoring its level of fatigue.
- 39
- 32
- Monocular Head Pose Estimation31
3D rigid head pose is estimated combining Active Appearance Models (AAM) with Pose from Orthography and Scaling with ITerations (POSIT) and a statistical anthropometric 3D model is used as reference.
- Iterative multistep explicit camera calibration31
The proposed methodology has the advantage that it can be used with monoplane calibration data with no restrictions for the pose geometry of the camera, and makes this approach independent of the accuracy of the horizontal scale factor.
- 29
- Generative face alignment through 2.5D active appearance models29
A 2.5D AAM that combines a 3D metric Point Distribution Model (PDM) and a 2D appearance model whose control points are defined by a full perspective projection of the PDM, which can outperform both the 2D+3D combined models and the2D standard methods.
- 29
- Tracking Pedestrians Under Occlusion Using Multiple Cameras29
A integrated solution to track multiple non-rigid objects (pedestrians) in a multiple cameras system with ground-plane trajectory prediction and occlusion modelling and the resulting system is able to maintain the tracking of pedestrians before, during and after Occlusion.
- 28
- Discriminative Bayesian Active Shape Models28
This work presents a simple and very efficient solution to align facial parts in unseen images that improves on the state-of-the-art while using the same local detectors and shows that MOSSE filters further improve on these results.
- A modified Casteljau algorithm to solve interpolation problems on Stiefel manifolds26
A new method to generate smooth interpolating curves on Stiefel manifolds is proposed from a modification of the geometric Casteljau algorithm on manifolds and is based on successive quasi-geodesic interpolation.
- 25
- Semi-intrinsic Mean Shift on Riemannian Manifolds24
This work proposes a novel paradigm to intrinsically reformulate the mean shift on general Riemannian manifolds by embedding the Riem Mannian manifold into a Reproducing Kernel Hilbert Space by using a general and mathematically well-founded RiemANNian kernel function.
- Strength in Diversity: Multi-Branch Representation Learning for Vehicle Re-Identification22
This paper presents an efficient and lightweight multi-branch deep architecture to improve vehicle reidentification (V-ReID), and argues that specialized loss-branch-splitting helps to improve re-identification tasks by generating specialized re-identification features.
- 22
- 21
- 21
- 19
- Lifting Object Detection Datasets into 3D18
This paper proposes to bypass previous solutions such as 3D scanning or manual design, that scale poorly, and instead populate object category detection datasets semi-automatically with dense, per-object 3D reconstructions, bootstrapped from class labels, figure-ground segmentations and a small set of keypoint annotations.
- Optical Normal Flow Estimation on Log-polar Images. A Solution for Real-Time Binocular Vision18
One solution for the problem based on log-polar images, including the algorithm implementation on digital signal processing hardware, is described, developed to be used with an active vision system to pursue moving objects, assuming that the cameras could also move.
- 17
- 17
- 16
- 16
- 14
- 13
- 13
- Locating Facial Features Using an Anthropometric Face Model for Determining the Gaze of Faces in Image Sequences13
A framework that combines a robust facial features location with an elliptical face modelling to measure user's intention and point of attention is presented and experimental results using real images sequences demonstrates the accuracy and robustness of the proposed solution.
- Maintaining the relative positions and orientations of multiple robots using vision13
This system has a mixed control structure, where both a central coordinator and each robot has a high level of autonomy and the relative locations of the robots are estimated using vision.
- Monoplanar Camera Calibration Iterative Multi-Step Approach13
A new method for robotics camera calibration based on a linear computation and using a coplanar group of calibration points is presented, which suppresses the constraint presented by the RAC calibration model.
- Stereoscopic Depth Perception Using a Model Based on the Primary Visual Cortex12
This work describes an approach inspired by the primary visual cortex using the stimulus response of the receptive field profiles of binocular cells for disparity computation and proposes a suitable model to consistently represent the complex cells by computing the wide bandwidths of the cortical cells.
- 12
- 12
- Real-time human activity monitoring exploring multiple vision sensors12
The monitoring of human activity in an indoor environment through the use of multiple vision sensors, made up of three cameras, enables the creation of a record based on the type of activity.
- Improving 3D Active Visual Tracking12
This paper describes the results of the analysis of a binocular tracking system and proposes a methodology for such an analysis (and resulting enhancements) based on techniques from control theory.
- 11
- 11
- FACIAL EXPRESSION RECOGNITION USING ACTIVE APPEARANCE MODELS11
A framework for automatic facial expression recognition combining Active Appearance Model (AAM) and Linear Discriminant Analysis (LDA) is proposed and the performed classification is based on malahanobis distance.
- FACIAL EXPRESSION RECOGNITION USING ACTIVE APPEARANCE MODELS11
A framework for automatic facial expression recognition combining Active Appearance Model (AAM) and Linear Discriminant Analysis (LDA) is proposed and the performed classification is based on malahanobis distance.
- Integration of information from several vision systems for a common task of surveillance11
The integration of vision information obtained by a fixed camera and the information obtaining by a binocular active vision system is described to achieve a higher degree of robustness in detecting and tracking intruders.
- Face Alignment Through 2.5D Active Appearance Models10
This work addresses the fitting of 3D deformable face models from a single view through 2.5D Active Appearance Models (AAM), using a 3D metric Point Distribution Model and a 2D appearance model whose control points are defined by full perspective projections of the PDM.
- Visual Behaviors for Real-Time Control of a Binocular Active Vision System10
This paper discusses a number of issues related with the implementation of a real-time control architecture and describes the architecture it is using with camera heads, and claims that higher levels of autonomy and integration can be obtained by designing the system architecture based on the concept of purposive behavior.
- 10
- 9
- 9
- Let the Shape Speak - Discriminative Face Alignment using Conjugate Priors9
This work presents a novel Bayesian global optimization strategy, where the prior is used to encode the dynamic transitions of the PDM parameters and model the prior distribution of the data as being Gaussian.
- 9
- 8
- Robust Segmentation Process to Detect Incidents on Highways8
This segmentation process is based on background subtraction and uses an efficient background model initialisation and update to work 24/7 and a cross-correlation based shadow detection is used for minimising ghosts.
- 8
- 7
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- Exploring Multi-Loss Learning for Multi-View Fine-Grained Vehicle Classification5
This work focuses on solving vehicle make and model classification by developing a fine-grained vehicle classification system that exploits the multi-camera composition of the system by powering a convolutional neural network with multiple views of the vehicle.
- Exploring Workout Repetition Counting and Validation Through Deep Learning5
This work proposes a method for workout repetition counting and validation based on a set of skeleton-based and deep semantic features that are obtained from a 2D human pose estimation network.
- 5
- A region covariance embedded in a particle filter for multi-objects tracking5
The proposed architecture is capable of tracking multiple objects even in the presence of periods of full occlusions using a simple and efficient solution for group handling and occlusion reasoning.
- 5
- Background Modelling on Tensor Field for Foreground Segmentation4
A novel Kmeans algorithm based on a new family of metrics, called Log-Euclidean, is proposed in order to speed up the process, while conserving the same theoretical properties, in the Affine-Invariant framework.
- Estimating Vehicle Velocity Using Image Profiles on Rectified Images4
A technique to estimate vehicle velocity using intensity profiles does not require background estimation or even the identification and tracking of individual vehicles, but requires the estimation of virtual images that represent a bird eye view of the scenario.
- ESTIMATING VEHICLE VELOCITY USING RECTIFIED IMAGES4
A technique to estimate vehicles velocity, using rectified images that represent a top view of the highway, and to rectify image sequences captured by uncalibrated cameras, which automatically estimates two vanishing points using lines from the image plane.
- A Framework for Wrong Way Driver Detection Using Optical Flow4
The proposed solution to detect wrong way drivers on highways is based on three main stages: Learning, Detection and Validation, which has proven to be quite robust in terms of different weather conditions, illumination and image quality.
- 4
- 3
- Bio-inspired Binocular Disparity with Position-Shift Receptive Field3
A bio-inspired method to calculate binocular disparity based on the energy model for depth perception of real world images is described and implemented and the performance of the proposed method is evaluated in what concerns the algorithm computational cost, root-mean-square error and percentage of wrongly matched pixels.
- 3
- 3
- A Distributed Control Network for Sensory Robotics3
The development of a robotic system used for the evaluation of sensory integration techniques and real-time path control strategies, featuring vision, force, tactile, range and proximity sensors is described.
- 2
- Gradient Shape Model2
The Gradient Shape Model (GSM) is introduced, a novel approach that is able to align a similar CLM model without the need for any convolutions at all and allows a substantially reduction in the complexity/dimensionality of the data term, making it possible to compute a denser, more accurate, regression step per cascade level.
- 2
- Kernel Based Multi-object Tracking Using Gabor Functions Embedded in a Region Covariance Matrix2
The proposed architecture is capable of tracking multiple objects even in the presence of periods of full occlusion, and analyzed the advantages of using Gabor functions as features and embedded them in the RCMs to get a more accurate descriptor.
- 2
- Simultaneous Identity and Expression Recognition Using Face Geometry on Low Dimensional Manifolds2
The proposed solution starts by extracting face geometry from input images using Active Appearance Models (AAM) and low dimensional manifolds were derived using Laplacian EigenMaps resulting in two types of manifolds, one for model identity and the other for expression.
- 2
- Combination of several vision sensors for interpretation of human actions2
How the combination of several vision sensors can be used to track multiple human targets in a typical surveillance situation and the creation of a record based on the type of activity is described.
- 2
- 1
- Biologically inspired computational modeling of motion based on middle temporal area1
A bio-inspired algorithm for motion computation based on V1 (Primary Visual Cortex) andMT (Middle Temporal Area) cells and a normalization and a pooling to find the most active neurons for motion detection are described.
- 1
- 1
- 1
- TOWARDS GENERIC FITTING USING DISCRIMINATIVE ACTIVE APPEARANCE MODELS EMBEDDED ON A RIEMANNIAN MANIFOLD1
The proposed discriminative fitting method found candidates to solutions based on a mean-shift algorithm, followed by an unsupervised clustering technique used to locate and group the candidates, demonstrating the improvement in performance.
- –
- –
- Tractable probabilistic models and computational complexity–
The framework that is put forward builds on a construction of a circuit representation of a dataset and on the definition of problems that take that circuit as an input such that the answers of the problems correspond to values of interest.
- –
- –
- –
- Wheels for Staircases–
The objective is to design innovative robot wheels capable of rolling on staircases, without sliding and without bouncing, the first step to reach the ultimate goal of building wheelchairs capable to overcome the obstacles imposed by staircases on people with limited mobility.
- –
- –
- –
- –
- –
- DYNAMIC GLOBAL OPTIMIZATION FRAMEWORK FOR REAL-TIME TRACKING–
This work presents an extension to the new Dynamic Hungarian Algorithm that achieves this effect, and shows tracking results in such different conditions as the tracking of humans and vehicles, in different scenes, using the same set of parameters for the tracker.
- SHADOW MODELING AND DETECTION FOR ROBUST FOREGROUND SEGMENTATION IN HIGHWAY SCENARIOS–
A method to automatically model and detect shadows on highway surveillance scenarios using a cascade of two classifiers and the advantages of applying spatial contextualization methods to the weak and strong classifiers is presented.
- –
- Active vision for autonomous systems–
Two systems with potential application to surveillance are described, one of which involves the integration of an active vision system in a mobile platform and the other can track non-rigid objects in real-time by using differential flow.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.