This work investigates classification of emotions from full-body movements by using a novel Convolutional Neural Network-based architecture. The model is composed of two shallow networks processing in parallel when the 8-bit RGB images obtained from time intervals of 3D-positional data are the inputs. One network performs a coarse-grained modelling in the time domain while the other one applies a fine-grained modelling. We show that combining different temporal scales into a single architecture improves the classification results of a dataset composed of short excerpts of the performances of professional dancers who interpreted four affective states: anger, happiness, sadness, and insecurity. Additionally, we investigate the effect of data chunk duration, overlapping, the size of the input images and the contribution of several data augmentation strategies for our proposed method. Better recognition results were obtained when the duration of a data chunk was longer, and this was further improved by applying balanced data augmentation. Moreover, we test our method on other existing motion capture datasets and compare the results with prior art. In all experiments, our results surpassed the state-of-the-art approaches, showing that this method generalizes across diverse settings and contexts.

No abstract

This work investigates classification of emotions from MoCap full-body data by using Convolutional Neural Networks (CNN). Rather than addressing regular day to day activities, we focus on a more complex type of full-body movement -dance. For this purpose, a new dataset was created which contains short excerpts of the performances of professional dancers who interpreted four emotional states: anger, happiness, sadness, and insecurity. Fourteen minutes of motion capture data are used to explore different CNN architectures and data representations. The results of the four-class classification task are up to 0.79 (F1 score) on test data of other performances by the same dancers. Hence, through deep learning, this paper proposes a novel and effective method of emotion classification which can be exploited in affective interfaces.