arXiv Computer Vision research summaries for April 18, 2024.

Today's Research Themes (AI-Generated):

• CT2X-GAN synthesizes realistic X-ray images from CT scans with superior quality to existing methods, improving safety by reducing radiation exposure.

• FreeDiff refines image editing with diffusion models using frequency truncation, offering a versatile tool without complex modifications or task limitations.

• AG-NeRF presents an end-to-end method for efficient, high-fidelity rendering of large-scale outdoor scenes from multi-height images.

• The proposed symmetric regressor improves MRI-based Parkinson's disease assessment by predicting DAT uptake amounts for both right and left striata simultaneously.

• Sketch-guided image inpainting with PDDP significantly outperforms existing approaches and provides greater user control over object shape and pose.