Table 1 - uploaded by Roberto Valenti
Content may be subject to copyright.
Source publication
Head pose and eye location estimation are two closely related issues which refer to similar application areas. In recent years, these problems have been studied individually in numerous works in the literature. Previous research shows that cylindrical head models and isophote based schemes provide satisfactory precision in head pose and eye locatio...
Context in source publication
Similar publications
Eye Localization is a preprocessing step for operations such as orientation correction and scaling required in face recognition problem. The success of facial feature analysis and face recognition system depends on eye position detected and the other facial features estimated on this. This paper presents a hybrid approach for eye localization in a...
Citations
... Later, relative positions of facial landmarks were applied. A wide variety of ways were proposed to extract and represent facial features [12][13][14][15][16][17]. Using landmarks is computationally efficient due to simple feature representations, and the detected landmarks are robust to transformation. ...
It is a routine procedure to use Iterative Closest Point (ICP) algorithm to correct motion in medical imaging of the head. However, the basic assumption of ICP is the rigid body motion, and it is not strictly valid when there is non-rigid motion (facial expression). In this study, we propose a novel method to reduce the adverse effects of facial expressions on head motion estimation. First, we design a network named DFG-Net to generate a deformation field on the RGB domain and compute a confidence map. We then extend the original ICP by embedding the confidence map representing each point’s degree of non-rigid motion. Compared to other techniques, our proposed method eliminates the negative effects of facial expression while making the most of pose information. The experiments demonstrate that our approach reduces the position error by 19.23% and orientation error by 36.37% compared to the original ICP algorithm. They also show that our method is robust to variations of expressions.
... All frames Frames with outliers (Cheung et al., 2015) 2,69 4,53 5,48 4,23 Vicente 2015 (Vicente et al., 2015) 3,20 4,30 6,20 4,57 Sung 2008 3,10 5,40 5,60 4,70 Kumano 2009(Kumano et al., 2009 2,90 7,10 4,20 4,73 Guo 2012 (Guo et al., 2012) 5,30 4,90 4,80 5, 00 Valenti 2009(Valenti et al., 2009 4,20 6,60 6,40 5,73 La Cascia 2000(La Cascia et al., 2000 9,80 3,30 6,10 6,40 tracking conditions. The accuracy of the 3D head tracker is of 1.09 o , and it is incremented to only 1.84 • when the tracking is very noisy, in those 23% of the frames that contain at least one outlier. ...
Head pose estimation (HPE) is currently a growing research field, mainly because of the proliferation of human–computer interfaces (HCI) in the last decade. It offers a wide variety of applications, including human behavior analysis, driver assistance systems or gaze estimation systems. This article aims to contribute to the development of robust and accurate HPE methods based on 2D tracking of the face, enhancing performance of both 2D point tracking and 3D pose estimation. We start with a baseline method for pose estimation based on POSIT algorithm. A novel weighted variant of POSIT is then proposed, together with a methodology to estimate weights for the 2D–3D point correspondences. Further, outlier detection and correction methods are also proposed in order to enhance both point tracking and pose estimation. With the aim of achieving a wider impact, the problem is addressed using a global approach: all the methods proposed are generalizable to any kind of object for which an approximate 3D model is available. These methods have been evaluated for the specific task of HPE using two different head pose video databases; a recently published one that reflects the expected performance of the system in current technological conditions, and an older one that allows an extensive comparison with state-of-the-art HPE methods. Results show that the proposed enhancements improve the accuracy of both 2D facial point tracking and 3D HPE, with respect to the implemented baseline method, by over 15% in normal tracking conditions and over 30% in noisy tracking conditions. Moreover, the proposed HPE system outperforms the state of the art on the two databases.
... For this tracker we use a cylindrical head pose tracker algorithm based on the Lukas-Kanade optical flow method [54]. The integration of the cylindrical head pose tracker and the isophote based eye tracker (described in section 2.1) is outside the scope of this manuscript, refer to [49] for detailed description. Next we give a short description of the initialization step of the cylindrical head pose tracker. ...
... • the second aspect is the detection error -this error has two aspects itself -the errors committed by the eye center locator and the errors committed by the head tracker. The used system for eye center location [48] claims an accuracy close to 100% for the eye center being located within 5% of the interocular distance and the head tracker accuracy is discussed in [49]. These errors will expand even further the uncertainty window associated with the maximum shift of the eye in the images obtained through the low resolution web camera. ...
... • head tracking errors -there are errors associated with the head tracker as well. The head tracker accuracy is discussed in [49]. Since we incorporate the knowledge for the current position of the head in 3D space into our final decision on point of visual gaze it is clear that these errors contribute to the final device error. ...
This manuscript investigates and proposes a visual gaze tracker that tackles the problem using only an ordinary web camera and no prior knowledge in any sense (scene set-up, camera intrinsic and/or extrinsic parameters). The tracker we propose is based on the observation that our desire to grant the freedom of natural head movement to the user requires 3D modeling of the scene set-up. Although, using a single low resolution web camera bounds us in dimensions (no depth can be recovered), we propose ways to cope with this drawback and model the scene in front of the user. We tackle this three-dimensional problem by realizing that it can be viewed as series of two-dimensional special cases. Then, we propose a procedure that treats each movement of the user's head as a special two-dimensional case, hence reducing the complexity of the problem back to two dimensions. Furthermore, the proposed tracker is calibration free and discards this tedious part of all previously mentioned trackers. Experimental results show that the proposed tracker achieves good results, given the restrictions on it. We can report that the tracker commits a mean error of (56.95, 70.82) pixels in x and y direction, respectively, when the user's head is as static as possible (no chin-rests are used). Furthermore, we can report that the proposed tracker commits a mean error of (87.18, 103.86) pixels in x and y direction, respectively, under natural head movement.
... The iris region is located in [29] using intensity differences between the center region and its neighboring regions using a contrast operator, and the detection accuracy is further enhanced using a Kalman tracker. Valenti et al. [30,31,32,33] propose a technique based on isophote curvatives and a voting process for real-time eye center detection with high accuracy, while a modified Mean Shift procedure is used for tracking. The desired features of the eyes can be enhanced using several filter responses [34]. ...
... Although their method does not yield superior accuracy for precise detection (ϵ ≤ 0.05), it is reported to have the best average performance over all values of ϵ. The method of Valenti et al. [30,31,32,33] which relies on isophote curvatives, reports a considerable improvement in accurately locating eye centers when compared to other state-of-the-art-methods. For databases where the accuracy results were not explicitly reported in their work (i.e. for GTAV and MUCT databases), a publicly available implementation 1 of their basic algorithm using the optimal parameters (obtained through grid search), was used for comparison. ...
The localization of eye centers and tracking of gaze constitutes an integral component of many human–computer interaction applications. A number of constraints including intrusiveness, mobility, robustness and high-price of eye tracking systems have hindered the way of the integration of eye trackers in everyday applications. Several ‘passive’ systems based on a single camera have been lately proposed in the literature, exhibiting though subordinate precision compared to the commercial, hardware-based eye tracking devices. In this paper we introduce an automatic, non-intrusive method for precise eye center localization in low resolution images, acquired from single low-cost cameras. To this end, the proposed system uses color information to derive a novel eye map that emphasizes the iris area and a radial symmetry transform which operates both on the original eye images and the eye map. The performance of the proposed method is extensively evaluated on four publicly available databases containing low resolution images and videos. Experimental results demonstrate great accuracy in challenging cases and resilience to pose and illumination variations, achieving significant improvement over existing methods.
... There have been numerous approaches to tackle the head pose estimation problem. The temporal information from face video sequence could improve the head pose estimation [6], but the applications of head pose estimation from a single 2D image are more extensive. Furthermore it is still difficult. ...
Head pose estimation from single 2D images has been considered as an important and challenging research task in computer vision. This paper presents a novel head pose estimation method which utilizes the shape model of the Basel face model and five fiducial points in faces. It adjusts shape deformation according to Laplace distribution to afford the shape variation across different persons. A new matching method based on PSO (particle swarm optimization) algorithm is applied both to reduce the time cost of shape reconstruction and to achieve higher accuracy than traditional optimization methods. In order to objectively evaluate accuracy, we proposed a new way to compute the pose estimation errors. Experiments on the BFM-synthetic database, the BU-3DFE database, the CUbiC FacePix database, the CMU PIE face database, and the CAS-PEAL-R1 database show that the proposed method is robust, accurate, and computationally efficient.
... Appearance-based works have generally made use of classifiers such as Support Vector Machines (SVMs)[4]and regression-based techniques such as Support Vector Regression (SVR)[5,6,7]. Hybrid approaches are a combination of model-based and appearance-based approaches[8,9,10]. Each category of approaches has its limits and constraints. ...
... This paper focuses on the feature descriptor. Eye images can be gotten using algorithms15161718 The feature of Cooccurrence Matrix of Oriented Gradients (CMOG) consists of two parts. The first part is HOG descriptor, and the second part is cooccurrence matrix encoded with gradient information. ...
Accurate detection of driver's eye state by computer vision is critical to driver drowsiness monitoring. The histogram of oriented gradients (HOG) is commonly used as descriptive feature of eye image for state classification. However, HOG often suffers from the limit of local gradient information. This paper proposes a new HOG-like feature of eye image, called cooccurrence matrix of oriented gradients (CMOG), for the purpose of more effectively classifying the eye state. By introducing the cooccurrence matrix, the CMOG enhances the ability of describing global gradient information of eye images. The ZJU eye blink database is used as the baseline images for performance comparison. The classification results show that the accuracy of CMOG reaches up to 95.9% in comparison with 91.9% by HOG under this database.
... We give more insights about the method comparatively with our previous work and extend it to two degrees of freedom. Besides using public datasets (FacePix, CMU PIE and Boston University datasets [2][3][4]), we have also used web-cam captures in order to cover more situations such as completely occluded eye. Sample captures are available here: www.lifl.fr/~dahmane/VIDEOS. ...
... Also, we tested our approach on the CMU PIE dataset [3] which gives more variability in term of illumination and expressions (e.g., eyes closed or smile). In addition to image datasets, we tested the video sequences of the Boston University (BU) dataset [4]. In BU dataset, subjects are doing free movements including yaw and roll variations. ...
... We also test on videos as we aim to use the solution in real environment for having real pose-related feedback. We test our method on the video sequences of the Boston University head pose dataset [4]. We recall that the inclination of the symmetry axis corresponds to the roll angle in case of frontal view. ...
This paper addresses the problem of head pose estimation in order to infer non-intrusive feedback from users about gaze attention. The proposed approach exploits the bilateral symmetry of the face. Size and orientation of the symmetrical area of the face is used to estimate roll and yaw poses by the mean of decision tree model. The approach does not need the location of interest points on face and presents robustness to partial occlusions. Tests were performed on different datasets (FacePix, CMU PIE, Boston University) and our approach coped with variability in illumination and expressions. Results demonstrate that the changes in the size of the regions that contain a bilateral symmetry provide accurate pose estimation.
... In the pre-processing stage we extracted and registered the faces from the video frames to allow precise low level feature extraction. We used the eye tracker developed by [26] ...
... Eye detection and tracking is a fundamental task in computer vision, with broad applications in humancomputer interaction, behavioral analysis, and computer graphics, such as monitoring human vigilance [1], gazecontingent smart graphics [2], and assisting people with disability. Various methods for eye detection and tracking are reported in the literature and can be classified into three categories: Appearance based methods [3], Feature based methods [4], and template based methods [5, 6, and 7].Appearance based method detect eyes based on their photometric appearance and require a large amount of data for training the classifiers (e. g. neural network or the support vector machine). Feature-based methods explore the characteristics of the human eye to identify a set of distinctive features around the eyes. ...
