English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In the area of human fixation prediction, dozens of computational saliencymodels are proposed to reveal certain saliency characteristics under differentassumptions and definitions. As a result, saliency model benchmarking oftenrequires several evaluation metrics to simultaneously assess saliency modelsfrom multiple perspectives. However, most computational metrics are notdesigned to directly measure the perceptual similarity of saliency maps so thatthe evaluation results may be sometimes inconsistent with the subjectiveimpression. To address this problem, this paper first conducts extensivesubjective tests to find out how the visual similarities between saliency mapsare perceived by humans. Based on the crowdsourced data collected in thesetests, we conclude several key factors in assessing saliency maps and quantizethe performance of existing metrics. Inspired by these factors, we propose tolearn a saliency evaluation metric based on a two-stream convolutional neuralnetwork using crowdsourced perceptual judgements. Specifically, the relativesaliency score of each pair from the crowdsourced data is utilized toregularize the network during the training process. By capturing the keyfactors shared by various subjects in comparing saliency maps, the learnedmetric better aligns with human perception of saliency maps, making it a goodcomplement to the existing metrics. Experimental results validate that thelearned metric can be generalized to the comparisons of saliency maps from newimages, new datasets, new models and synthetic data. Due to the effectivenessof the learned metric, it also can be used to facilitate the development of newmodels for fixation prediction.

Learning a Saliency Evaluation Metric Using Crowdsourced Perceptual Judgments