Marek Kowalski

Deep Alignment Network: A Convolutional Neural Network for Robust Face Alignment

In this paper, we propose Deep Alignment Network (DAN), a robust face alignment method based on a deep neural network architecture. DAN consists of multiple stages, where each stage improves the locations of the facial landmarks estimated by the previous stage. Our method uses entire face images at all stages, contrary to the recently proposed face alignment methods that rely on local patches. This is possible thanks to the use of landmark heatmaps which provide visual information about landmark locations estimated at the previous stages of the algorithm. The use of entire face images rather than patches allows DAN to handle face images with large variation in head pose and difficult initializations. An extensive evaluation on two publicly available datasets shows that DAN reduces the state-of-the-art failure rate by up to 70%. Our method has also been submitted for evaluation as part of the Menpo challenge.

Live Scan3D: A Fast and Inexpensive 3D Data Acquisition System for Multiple Kinect v2 Sensors

LiveScan3D is a free, open source system for live, 3D data acquisition using multiple Kinect v2 sensors. It allows the user to place any number of sensors in any physical configuration and start gathering data at real time speed. The freedom of placing the sensors in any configuration allows for many possible acquisition scenarios such as: capturing a single object from many viewpoints or creating 3D panoramas with multiple devices located close to each other. Thanks to the off-the-shelf Kinect v2 sensor the system is both accurate and inexpensive, opening 3D acquisition up to more recipients. In the paper we describe our system with the algorithms it is using and show its effectiveness in multiple scenarios including head shape reconstruction and 3D reconstruction of dynamic scenes.

Using a Probabilistic Neural Network for lip-based biometric verification

In classical recognition techniques only raw features of objects are employed. Our approach allows use the composed features so called Sim coefficients and landmarks which determine the area where biometric features should be searched. Biometric composed features are associated with appropriate similarity coefficients. Such approach brings significant advantages recognition level of objects is higher compared to method based on the raw data. In this paper, a novel and effective lip-based biometric recognition approach with the Probabilistic Neural Network (PNN) is proposed. Lip based recognition has been less developed than the recognition of other human physical attributes such as the fingerprint, voice patterns, blood vessel patterns, or the face. For this reason, achieved results on this field are still improved and new recognition techniques are searched. Results achieved by PNN were improved by the Particle Swarm Optimization (PSO) technique.In the first step, lip area is restricted to a Region Of Interest (ROI) and in the second step, features extracted from ROI are specifically modeled by dedicated image processing algorithms. Extracted lip features are then an input data of neural network. All experiments were confirmed in the ten-fold cross validation fashion on three diverse datasets, Multi-PIE Face Dataset, PUT database and our own faces dataset. Obtained in researches result show that proposed approach achieves an average classification accuracy of 86.95%, 87.14%, and 87.26%, on these three datasets, respectively. Announced results were verified in the comparative studies and confirm the efficacy of the proposed lip based biometrics learned by PSO technique. We proposed a novel biometric system based on lip geometrical features measurements.Each lip feature is paired with a similarity measure and form a composed feature.The set of the most discriminative lip composed features is determined.Probabilistic Neural Network classifier is used for lip verification. Display Omitted

Face Alignment Using K-Cluster Regression Forests With Weighted Splitting

In this letter, we present a face alignment pipeline based on two novel methods: weighted splitting for K-cluster Regression Forests (KRF) and three-dimensional Affine Pose Regression (3D-APR) for face shape initialization. Our face alignment method is based on the Local Binary Feature (LBF) framework, where instead of standard regression forests and pixel difference features used in the original method, we use our K-Cluster Regression Forests with Weighted Splitting (KRFWS) and Pyramid Histogram of Oriented Gradients (PHOG) features. We also use KRFWS to perform APR and 3D-APR, which intend to improve the face shape initialization. APR applies a rigid 2-D transform to the initial face shape that compensates for inaccuracy in the initial face location, size, and in-plane rotation. 3D-APR estimates the parameters of a 3-D transform that additionally compensates for out-of-plane rotation. The resulting pipeline, consisting of APR and 3D-APR followed by face alignment, shows an improvement of 20% over standard LBF on the challenging Intelligent Behaviour Understanding Group (IBUG) dataset, and state-of-the-art accuracy on the entire 300-W dataset.

3D face data acquisition and modelling based on an RGBD camera matrix

This paper describes a novel system for building morphable 3D head models. In contrast to most of the previous approaches that need several seconds to capture each scan, we acquire the data using a matrix of calibrated RGBD cameras, enabling real time face scanning. We localize the face and its 68 characteristic points on an orthogonal projection image, and use the detected points to align multiple scans. We use a Delaunay triangulation of the 68 characteristic points to obtain dense head shapes with point to point correspondence across all 3D head shapes. In the last step we create a morphable model in a way that is similar to the original procedure by Blanz and Vetter. We demonstrate the functionality of our model, created on just five people, in a real-time application. The novelty of this article lies mostly in the method of defining correspondences of the characteristic points in 3D, that leads to a realistic three-dimensional model and blendshapes.

Evaluation of active appearance models in varying background conditions

In this paper we present an evaluation of the chosen versions of Active Appearance Models (AAM) in varying background conditions. Algorithms were tested on a subset of the CMU PIE database and chosen background im- ages. Our experiments prove, that the accuracy of those methods is strictly correlated with the used background, where the differences in the success rate differ even up to 50%.

Webcam-based system for video-oculography

Video-oculography (VOG) is a tool providing diagnostic information about the progress of the diseases that cause regression of the vergence eye movements, such as Parkinsons disease (PD). The majority of the existing systems are based on sophisticated infra-red (IR) devices. In this study, the authors show that a webcam-based VOG system can provide similar accuracy to that of a head-mounted IR-based VOG system. They also prove that the authors’ iris localisation algorithm outperforms current state-of-the-art methods on the popular BioID dataset in terms of accuracy. The proposed system consists of a set of image processing algorithms: face detection, facial features localisation and iris localisation. They have performed examinations on patients suffering from PD using their system and a JAZZ-novo head-mounted device with IR sensor as reference. In the experiments, they have obtained a mean correlation of 0.841 between the results from their method and those from the JAZZ-novo. They have shown that the accuracy of their visual system is similar to the accuracy of IR head-mounted devices. In the future, they plan to extend their experiments to inexpensive high frame rate cameras which can potentially provide more diagnostic parameters.

Head Pose Tracking from RGBD Sensor Based on Direct Motion Estimation

We propose to use a state-of-the-art visual odometry technique for the purpose of head pose estimation. We demonstrate that with small adaptation this algorithm allows to achieve more accurate head pose estimation from an RGBD sensor than all the methods published to date. We also propose a novel methodology to automatically assess the accuracy of a tracking algorithm without the need to manually label or otherwise annotate each image in a test sequence.

Online 3D face reconstruction with Incremental Structure From Motion and a regressor cascade

In this paper we present a method for online 3D face reconstruction from a video sequence. The face landmarks in a given frame are detected and used to create a 3D shape estimate. The resulting 3D shape is an approximate, sparse representation of the subject’s face. Our reconstruction step is based on a revised version of incremental Structure From Motion, where we use a novel 4D subspace tracking procedure followed by scaled deflation against a vector of ones. Facial landmark detection is built upon a regressor cascade scheme where each subsequent regressor updates the initial shape obtained from the preceding frame.