arXiv Computer Vision research summaries for February 19, 2024.

Today's Research Themes (AI-Generated):

• Innovative method for updating preoperative 3D anatomical models during sinus surgery using intraoperative endoscopy video.

• Introduction of WildFake, a large-scale dataset for detecting and analyzing AI-generated images.

• Introduction of UnlearnCanvas, a dataset to benchmark the machine unlearning of artistic painting styles in diffusion models.

• ComFusion approach for personalized text-to-image generation integrating user images and textual scene descriptions.

• Language-guided image reflection separation framework employing cross-attention mechanism with contrastive learning.