GER, or Gait Emotion Recognition, is a novel method of recognizing human emotions based on a person's walking pattern. Researchers have developed a classifier network called STEP that uses a Spatial Temporal Graph Convolutional Network (ST-GCN) architecture to classify an individual's perceived emotion into one of four categories: happy, sad, angry, or neutral.
The STEP Network
The STEP network is trained on annotated real-world gait videos, as well as synthetic gaits generated using a network called STEP-Gen. STEP-Gen is built on an ST-GCN-based Conditional Variational Autoencoder (CVAE) and incorporates a novel push-pull regularization loss to generate more realistic gaits.
The combination of real-world and synthetic gait videos allows the network to learn affective features that improve the accuracy of the classification process. The network achieves an accuracy of 88% on the E-Gait dataset, which is significantly more accurate than previous methods.
The E-Gait Dataset
The E-Gait dataset consists of 4,227 human gaits that are annotated with perceived emotions. The dataset also includes thousands of synthetic gaits generated by the STEP-Gen network. Researchers can use this dataset to train and evaluate their own gait emotion recognition networks.
The Importance of GER
The ability to recognize emotion based on gait can have important applications in a variety of fields. For example, in healthcare, GER could be used to monitor the emotional state of patients with mobility impairments, such as those with Parkinson's disease.
In the field of robotics, GER could be used to create robots that can recognize and respond to human emotions. Robots with this capability could be used in a variety of settings, from elderly care facilities to customer service.
GER is a novel method for recognizing human emotions based on a person's walking pattern. The STEP network uses a combination of real-world and synthetic gait videos to achieve high accuracy in emotion recognition. The E-Gait dataset provides a valuable resource for researchers interested in developing their own gait emotion recognition networks. The potential applications of GER are vast, with potential uses in healthcare, robotics, and more.
