Abstract: In many real-world prediction tasks, the class labels contain information about the relative order between the labels that are not captured by commonly used loss functions such as ...
Abstract: Multimodal video sentiment analysis aims to integrate multiple modal information to analyze the opinions and attitudes of speakers. Most previous work focuses on exploring the semantic ...
Designing an Encoder for StyleGAN Image Manipulation (SIGGRAPH 2021) Recently, there has been a surge of diverse methods for performing image editing by employing pre-trained unconditional generators.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results