arXiv Computer Vision research summaries for April 13, 2024.

Today's Research Themes (AI-Generated):

• Efficient adaptation of Vision Transformers with head-level tuning and Taylor-expansion Importance Scores improves model performance.

• Multi-modal prompting paradigm enhances few-shot medical image classification with prompt engineering and feature distribution focused heads.

• Introduction of large-scale Meply dataset and CoSAM model for lymph node detection and segmentation in rectal cancer.

• MAProtoNet achieves state-of-the-art performance in brain tumor localization using interpretable prototypical networks.

• Pyramidal Neural Representation for Videos addresses spatial inconsistency and sets new benchmarks for video regression.