Crowdsourced mouse-tracking data from a custom viewing system approximates eye-tracking saliency maps for videos and is improved by a proposed deep neural network.
We hired participants on Sub- jectify.us crowdsourcing platform, showed them 10 videos and paid them $0.15 if they watched all videos
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Predicting video saliency using crowdsourced mouse-tracking data
Crowdsourced mouse-tracking data from a custom viewing system approximates eye-tracking saliency maps for videos and is improved by a proposed deep neural network.