Contents shortcut Main menu shortcut
Life at CNU

CNU Students Win 1st Prize in International Emotion Recognition Competition

작성자대외협력과 작성일2019.08.27 17:01 조회79

CNU Students of the Department of Electrical, Electronic Communication & Computer Engineering won first prizes in two of three sections at the 2019 EmotiW (Emotion Recognition in the Wild) Competition.

The competition was organized by the 2019 of Association for Computing Machinery (ACM) International Conference on Multimodal Interaction 2019 and was divided into three main categories: ▲ GC (Group-level Cohesion Prediction) ▲ EW (Engagement prediction in the Wild) and ▲ AV (Audio-video-based Emotion Recognition).

The CNU “SML team”, featuring MA student Dang Xuan Tien and PhD student Vo Thanh Hung won the first prize in the ‘Group-level Cohesion Prediction’ (GC). This team’s advising professors are Kim Soo-hyung, Lee Guee-sang, and Yang Hyung-jeong

The SML team won the award by developing a Group Emotion to Group Cohesion (GE2GC) method that predicts group-level cohesion through a multi-stream hybrid network using the latest CNN models such as InceptionV3 and ResNet50 as feature extractors.

In addition, another research team of MA students (Huynh Van Thong, Vo Thi Tuong Vi, and Pham Van Cuong) won the first prize in the ‘Engagement prediction in the Wild’ (EW) section, also under the guidance of professors Kim Soo-hyung, Lee Guee-sang, and Yang Hyung-jeong.

The team ranked first in the field by presenting emotion analysis technology to characterize eye movements through a state-of-the-art architecture combining LSTM (long short term memory) and with FC (fiber channels).

ICMI (The International Conference on Multimodal Interaction) is the world's leading conference centering on artificial intelligence and multimodal emotion recognition. It holds a forum every year for international discussion to come up with improvement measures and better methods in the following areas: △ Emotion Recognition △ Human-machine Interaction △ Interface design △ Technology for Multimodal System Development.