A research team, led by Professor Taehwan Kim (UNIST Artificial Intelligence Graduate School) announced that they have accomplished outstanding achievements at an international conference on artificial intelligence (AI) and computer vision.
The Machine Learning, Computer Vision, and Language Processing Laboratory, run by Professor Kim won the second place in the CVPR 2022 LOng-form VidEo Understanding (LOVEU) challenges through industry-academia collaboration with the student-led startup, Pyler. According to the research team, this conference is considered to be one of the most prominent AI events in computer vision and pattern recognition.
The competition challenges competitors to develop a comprehensive AI model that is capable of understanding not only languages, but also videos and images, and even answering questions that require multiple steps. In this competition, the research team proposed a new context ground module attention mechanism for more effective feature mapping. In addition, they also performed the analysis over the number of buttons and ablation study of different step networks and video features. As a result, they achieved the overall 2nd place in the competition.
Figure 1. AI model architecture, developed by Professor Kim and his research team from the UNIST Artificial Intelligence Graduate School.
