Why is facial expression analysis in the wild challenging

摘要: In this paper, we discuss the challenges for facial expression analysis in wild. We studied problems exemplarily on Emotion Recognition Wild Challenge 2013 [3] dataset. performed extensive experiments dataset comparing different approaches face alignment, representation, and classification, as well human performance. It turns out that under close-to-real conditions, especially with co-occurring speech, it is hard even humans to assign emotion labels clips when only taking video into account. Our automatic classification achieved at best a correct rate of 29.81% test set using Gabor features linear support vector machines, which were trained web images. This result 7.06% better than official baseline, additionally incorporates time information.