Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation

Liu, Yunfei; Liu, Ruicong; Wang, Haofei; Lu, Feng

doi:10.1109/iccv48922.2021.00381

Cited by 36 publications

(32 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ADL * does not require target samples as our methods. In addition, we directly report the performance of other domain adaption methods from (Liu et al 2021) for reference.…”

Section: Domain Adaption Methodsmentioning

confidence: 99%

“…Wang et al (Wang et al 2019a) and Kellnhoder et al (Kellnhofer et al 2019) propose to use adversarial learning to align the features in the source and target domain. Liu et al (Liu et al 2021) propose an ensemble of networks that learn collaboratively with the guidance of outliers. These methods utilize data from target domain, which is not always user-friendly.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

PureGaze: Purifying Gaze Feature for Generalizable Gaze Estimation

Cheng

Bao

2022

AAAI

Self Cite

View full text Add to dashboard Cite

Gaze estimation methods learn eye gaze from facial features. However, among rich information in the facial image, real gaze-relevant features only correspond to subtle changes in eye region, while other gaze-irrelevant features like illumination, personal appearance and even facial expression may affect the learning in an unexpected way. This is a major reason why existing methods show significant performance degradation in cross-domain/dataset evaluation. In this paper, we tackle the cross-domain problem in gaze estimation. Different from common domain adaption methods, we propose a domain generalization method to improve the cross-domain performance without touching target samples. The domain generalization is realized by gaze feature purification. We eliminate gaze-irrelevant factors such as illumination and identity to improve the cross-domain performance. We design a plug-and-play self-adversarial framework for the gaze feature purification. The framework enhances not only our baseline but also existing gaze estimation methods directly and significantly. To the best of our knowledge, we are the first to propose domain generalization methods in gaze estimation. Our method achieves not only state-of-the-art performance among typical gaze estimation methods but also competitive results among domain adaption methods. The code is released in https://github.com/yihuacheng/PureGaze.

show abstract

“…ADL * does not require target samples as our methods. In addition, we directly report the performance of other domain adaption methods from (Liu et al 2021) for reference.…”

Section: Domain Adaption Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

PureGaze: Purifying Gaze Feature for Generalizable Gaze Estimation

Cheng

Bao

2022

AAAI

Self Cite

View full text Add to dashboard Cite

show abstract

“…Ref. [ 42 ] addressed the domain adaptation problem in gaze estimation. To generalize the gaze estimator to a new domain, ref.…”

Section: Related Workmentioning

confidence: 99%

“…To generalize the gaze estimator to a new domain, ref. [ 42 ] proposed a plug-and-play gaze adaptation framework, where outliers of the outputs of pre-trained models in the target domain are considered noisy labels, helping the collaborative learning process.…”

Section: Related Workmentioning

confidence: 99%

ArbGaze: Gaze Estimation from Arbitrary-Sized Low-Resolution Images

Kim

Chang

2022

Sensors

View full text Add to dashboard Cite

The goal of gaze estimation is to estimate a gaze vector from an image containing a face or eye(s). Most existing studies use pre-defined fixed-resolution images to estimate the gaze vector. However, images captured from in-the-wild environments may have various resolutions, and variation in resolution can degrade gaze estimation performance. To address this problem, a gaze estimation method from arbitrary-sized low-resolution images is proposed. The basic idea of the proposed method is to combine knowledge distillation and feature adaptation. Knowledge distillation helps the gaze estimator for arbitrary-sized images generate a feature map similar to that from a high-resolution image. Feature adaptation makes creating a feature map adaptive to various resolutions of an input image possible by using a low-resolution image and its scale information together. It is shown that combining these two ideas improves gaze estimation performance substantially in the ablation study. It is also demonstrated that the proposed method can be generalized to other popularly used gaze estimation models through experiments using various backbones.

show abstract

“…Interaction techniques aim to improve the user experience, which is vital for AR HMD devices. With the rise of gaze estimation accuracy [17,41,67], different gaze-based techniques have been explored, such as gaze dwelling and vergence eye movement.…”

Section: Gaze Interaction In Armentioning

confidence: 99%