As shown below, the inferred masks predicted by our segmentation model trained by the dataset appear similar to the ground truth masks. This is a heavily engineered, image-only, machine-learning-ready ...

We propose ModuSeg, a training-free weakly supervised semantic segmentation framework that explicitly decouples object discovery and semantic retrieval for high-quality pixel-level predictions. Weakly ...

These historic objects from the National Air and Space Museum's collection have been 3D scanned and 3D models produced in partnership with the Smithsonian Institution's Digitization Program Office.

Abstract: 3D Gaussian Splatting (3DGS) has emerged as a high-quality and efficient alternative to Neural Radiance Fields (NeRF), offering distinct advantages in scene representation and suitability ...

Abstract: Accurate segmentation of kidneys and renal lesions remains challenging due to anatomical variability, tumor heterogeneity, indistinct boundaries, and imaging inconsistencies. To address this ...