This project was a collaborative effort between myself, Marcus Lindelöf, Ruben Täpptorp, and Simon Rosengren. We focused on integrating the DeepFace framework into the social robot Epi to facilitate automatic facial expression recognition. We developed a custom Python pipeline using OpenCV to capture and analyze emotional data during a Wizard of Oz-style experiment. Our results demonstrated significant deviations between the robot's analysis and human perception, suggesting that social robotics requires multimodal approaches for accurate emotion detection. The repository for this project along with the associated instructions can be viewed below.
View Code on GitHubYou can view the complete descriptive article "Implementing Emotion Recognition in a Social Robot" below.
Open Article (PDF)