arXiv Computer Vision research summaries for April 07, 2024.

Today's Research Themes (AI-Generated):

• Enhancing low-light image performance for autonomous driving with multi-modal diffusion models.

• Improving foggy weather segmentations by decoupling defogging and semantic learning tasks.

• Real-time, memory-based optical flow estimation outperforms existing methods in video resolution adaptability.

• Joint 3D reconstruction of human-object interactions is refined with contact-based Transformers.

• Multi-level supervised learning reduces the demand for 3D-annotated samples in building reconstruction.