事前学習された深層学習モデルを活用し、顔画像から信頼できる感情特徴を抽出することで、感情分析タスクの精度を大幅に向上させることができる。
The author presents a two-step framework for multimodal emotion cause analysis, utilizing LLMs and GPT models to address challenges in capturing emotions in human conversations.
The author presents the 6th ABAW Competition focusing on understanding human emotions and behaviors through five challenges, emphasizing the importance of human-centered technologies.
The author explores methods for transferring emotions from resource-rich languages to low-resource languages, focusing on annotation projection and direct cross-lingual transfer. The study aims to improve emotion classification in low and moderate resource languages.