Analysis on Affect Recognition Methods in the Wild

Abstract

Affect recognition transition from laboratory-controlled to challenging in the wild conditions is an intense area of research, with a potentially long list of important application. Audio-visual modalities are significant contributors that provide rich contextual information from real world challenging corpora. These modalities can be explored for better discrimination of real world human emotions that are complex and compound. A comprehensive literature survey is carried out to identify the most relevant big databases and the way features are extracted and fused from visual and auditory data. The main focus is on recent state of Art research work using real-world corpora and the work comparing designed framework on controlled as well as in the wild data.