Few-Shot Personalized Saliency Prediction using Person Similarity based on Collaborative Multi-Output Gaussian Process Regression

Moroto, Yuya; Maeda, Kazuo; Ogawa, Takahiro; Haseyama, Miki

doi:10.1109/icip42928.2021.9506583

Cited by 2 publications

(11 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the perspective of the quantitative evaluation, we compare the proposed method with other methods in Table 1. The proposed method outperforms all comparative methods in all metrics, while the proposed method outperforms CoMOGPbased FPSP [20] in "CC" and "KLdiv". Thus, we confirm the effectiveness of the proposed method.…”

Section: Performance Evaluationmentioning

confidence: 90%

“…CoMOGP-based FPSP: A few-shot PSM prediction method using CoMOGP instead of Sec.2.3 in our method. Note that we trained the PSM prediction methods with only the common images selected by AIS [20].…”

Section: Experimental Settingsmentioning

confidence: 99%

“…Moreover, by comparing our method with Similarity-based FPSP [19], the effectiveness of introducing object information into the similarity of gaze tendency is confirmed. Next, we compare our method with CoMOGP-based FPSP [20] that can consider the semantic information. The proposed method outperforms CoMOGPbased FPSP [20] in most evaluation metrics, and the effectiveness of the use of structural information is confirmed.…”

Section: Performance Evaluationmentioning

confidence: 99%

“…In the previous study, adaptive image selection (AIS) has been proposed for selecting such images for calculating the similarities of gaze tendency between persons [19]. Furthermore, several previous methods have predicted PSMs of the target persons based on the similarities of gaze tendency between the target and other persons (hereafter, training persons) for whom eye tracking data are obtained enough by using images selected by the AIS [19,20]. It should be noted that the training persons are already prepared by the above PSM dataset.…”

Section: Introductionmentioning

confidence: 99%

“…It should be noted that the training persons are already prepared by the above PSM dataset. In [20], the collaborative multi-output Gaussian process regression (CoMOGP) [21] using visual information obtained from images, and predicted PSMs of the training persons is adopted for predicting the PSM with a limited amount of data. By using visual information, the CoMOGP-based method can take into account the semantic information of images.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Few-Shot Personalized Saliency Prediction with Similarity of Gaze Tendency Using Object-Based Structural Information

Moroto

Maeda

Ogawa

et al. 2022

2022 IEEE International Conference on Image Processing (ICIP)

Self Cite

View full text Add to dashboard Cite

This paper presents a few-shot personalized saliency prediction method with similarity of gaze tendency using object-based structural information. The personalized saliency maps (PSMs) that represent the individual visual attention can be used for analyzing the heterogeneity among personalized preferences, whereas general saliency maps ignore the individual differences. However, the PSM prediction is a difficult task since the acquisition of eye tracking data, which are needed for obtaining PSMs, gives persons a heavy burden. Then, for realizing PSM prediction with a limited amount of training data, the use of the similarity of gaze tendency between persons can be one effective way. It has been reported that human gazes are related to objects and their relative relationships that are semantic and structural information, and we focus on the integration of PSMs predicted for other persons by using object-based similarities of gaze tendency. The advantage in this paper is that we newly focus on similarities of gaze tendency for visually similar objects for solving the lack of eye tracking data by considering both semantic and structural information, simultaneously. By experimenting with the open dataset, the proposed method outperforms the state-of-the-art methods.

show abstract

Section: Performance Evaluationmentioning

confidence: 90%

Section: Experimental Settingsmentioning

confidence: 99%