Abstract

Given a single outdoor image, this paper proposes acollaborative learning approach for labeling it as eithersunny or cloudy. Never adequately addressed, this two-class classification problem is by no means trivial giventhe great variety of outdoor images. Our weather featurecombines special cues after properly encoding them intofeature vectors. They then work collaboratively in synergyunder a unified optimization framework that is aware of the presence (or absence) of a given weather cue during learning and classification. Extensive experiments and comparisons are performed to verify our method. We build a new weather image dataset consisting of 10K sunny and cloudy images, which is available online together with the executable.