Leveraging multiple cues for recognizing family photos

Wang, Xiaolong; Guo, Guodong; Merler, Michele; Codella, Noel; Rohith, M.; Smith, J. R.; Kambhamettu, Chandra

doi:10.1016/j.imavis.2016.07.006

Cited by 12 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The follow-up works in this field keep focusing on photo albums and consider a wider range of relations [27–29]. Additionally, they profit from the abundance of social network data by employing more powerful tools such as Deep Neural Networks (DNN) [30–32].…”

Section: Introductionmentioning

confidence: 99%

Identification of social relation within pedestrian dyads

Yücel

Zanlungo²,

Feliciani

et al. 2019

PLoS ONE

View full text Add to dashboard Cite

This study focuses on social pedestrian groups in public spaces and makes an effort to identify the type of social relation between the group members. As a first step for this identification problem, we focus on dyads (i.e. 2 people groups). Moreover, as a mutually exclusive categorization of social relations, we consider the domain-based approach of Bugental, which precisely corresponds to social relations of colleagues, couples, friends and families, and identify each dyad with one of those relations. For this purpose, we use anonymized trajectory data and derive a set of observables thereof, namely, inter-personal distance, group velocity, velocity difference and height difference. Subsequently, we use the probability density functions (pdf) of these observables as a tool to understand the nature of the relation between pedestrians. To that end, we propose different ways of using the pdfs. Namely, we introduce a probabilistic Bayesian approach and contrast it to a functional metric one and evaluate the performance of both methods with appropriate assessment measures. This study stands out as the first attempt to automatically recognize social relation between pedestrian groups. Additionally, in doing that it uses completely anonymous data and proves that social relation is still possible to recognize with a good accuracy without invading privacy. In particular, our findings indicate that significant recognition rates can be attained for certain categories and with certain methods. Specifically, we show that a very good recognition rate is achieved in distinguishing colleagues from leisure-oriented dyads (families, couples and friends), whereas the distinction between the leisure-oriented dyads results to be inherently harder, but still possible at reasonable rates, in particular if families are restricted to parent-child groups. In general, we establish that the Bayesian method outperforms the functional metric one due, probably, to the difficulty of the latter to learn observable pdfs from individual trajectories.

show abstract

Section: Introductionmentioning

confidence: 99%

Identification of social relation within pedestrian dyads

Yücel

Zanlungo²,

Feliciani

et al. 2019

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…As an independent application, specific frontalization techniques have also been proposed [9]. Another line of work pertains to 3D face reconstruction from photo collections [29,18,42] or a single image [19,50,40], where the latter have been successfully used for face normalization prior to recognition. While most of the methods apply the framework of aligning 3DMM with the 2D face landmarks [47,46,25] and conduct further refinement.…”

Section: Related Workmentioning

confidence: 99%

Reconstruction-Based Disentanglement for Pose-Invariant Face Recognition

Peng

Sohn

et al. 2017

2017 IEEE International Conference on Computer Vision (ICCV)

144

View full text Add to dashboard Cite

Deep neural networks (DNNs) trained on large-scale datasets have recently achieved impressive improvements in face recognition. But a persistent challenge remains to develop methods capable of handling large pose variations that are relatively under-represented in training data. This paper presents a method for learning a feature representation that is invariant to pose, without requiring extensive pose coverage in training data. We first propose to generate non-frontal views from a single frontal face, in order to increase the diversity of training data while preserving accurate facial details that are critical for identity discrimination. Our next contribution is to seek a rich embedding that encodes identity features, as well as non-identity ones such as pose and landmark locations. Finally, we propose a new feature reconstruction metric learning to explicitly disentangle identity and pose, by demanding alignment between the feature reconstructions through various combinations of identity and pose features, which is obtained from two images of the same subject. Experiments on both controlled and in-the-wild face datasets, such as MultiPIE, 300WLP and the profile view database CFP, show that our method consistently outperforms the state-of-the-art, especially on images with large head pose variations. 1

show abstract

“…Only well-fitted faces will be used to incrementally update the representation subspace and adapt the cascade of regressors for person-specific modeling Our goal is to learn a deep neural network that takes the fitting results as input and outputs a binary label to indicate correct or erroneous alignment. To connect the facial appearance and the fitted shape, a possible solution is to directly concatenate the vector of landmark coordinates to an intermediate fully connected layer [41,42]. However, we experienced very limited performance using this design in our experiments.…”

Section: Deep Fitting Evaluationmentioning

confidence: 99%

Track Facial Points in Unconstrained Videos

Peng

Huang

et al. 2016

Procedings of the British Machine Vision Conference 2016

View full text Add to dashboard Cite

Tracking Facial Points in unconstrained videos is challenging due to the non-rigid deformation that changes over time. In this paper, we propose to exploit incremental learning for person-specific alignment in wild conditions.Our approach takes advantage of part-based representation, as illustrated in Figure 1 and cascade regression for robust and efficient alignment on each frame. Unlike existing methods that usually rely on models trained offline, we incrementally update the representation subspace and the cascade of regressors in a unified framework to achieve personalized modeling.Blind model adaptation without correction would inevitably result in model drifting. How to effectively detect misalignment is still a challenging question that is seldom investigated. To address this issue, we propose a deep neural network for robust fitting evaluation to pick out well-aligned faces from misalignment. The architecture of the network is shown is Figure 2. These well-aligned faces are then used to incrementally update the representation subspace and fitting strategy for person-specific modeling on the fly. In summary, our work makes the following contributions:(1) We propose a novel approach for sequential face alignment. To the best of our knowl- edge, this is the first time that person-specific modeling is investigated to jointly learn the representation subspace and the fitting parameters in a unified framework.(2) The proposed part-based representation together with the cascade regression guarantees robust alignment in unconstrained conditions. More importantly, they are critical to efficiently construct personalized models for real-time or large-scale applications.(3) We propose to leverage deep neural networks for efficient and robust fitting evaluation. It significantly alleviates the drifting issue which would severely deteriorate learned models and inevitably lead to failure.To validate our approach, we provide a detailed experimental analysis of each component of our approach, as well as performance comparisons with existing approaches. Four image datasets (MultiPIE, LFPW, Helen, and AFLW), four video datasets (FGNET, ASLV, 300-VW, and YtbVW), and four state of the arts (RLMS, DRMF, IFA, and ESR) are employed to conduct the experiments. The results demonstrate that the proposed incremental learning can significantly improve the fitting accuracy with an affordable computational cost, especially in unconstrained videos with extensive variations.

show abstract

Leveraging multiple cues for recognizing family photos

Cited by 12 publications

References 19 publications

Identification of social relation within pedestrian dyads

Identification of social relation within pedestrian dyads

Reconstruction-Based Disentanglement for Pose-Invariant Face Recognition

Track Facial Points in Unconstrained Videos

Contact Info

Product

Resources

About