UnSAMv2: Self-Supervised Learning Enables Segment Anything at Any Granularity
cs.CV, cs.AI, cs.LG
Submitted: 2025-11-17
Updated: 2026-10-10
Project page: https://yujunwei04.github.io/UnSAMv2-Project-Page
Terminology
Sources
- ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depth
- SOHES: Self-supervised Open-world Hierarchical Entity Segmentation
- Self-supervised Learning of Hybrid Part-aware 3D Representations of 2D Gaussians and Superquadrics
- Unsupervised Semantic Segmentation by Distilling Feature Correspondences
- RayZer: A Self-supervised Large View Synthesis Model
- RefCut: Interactive Segmentation with Reference Guidance
- DINOv2: Learning Robust Visual Features without Supervision
- SAM 2: Segment Anything in Images and Videos
- Localizing Objects with Self-Supervised Transformers and no Labels
- DINOv3
- Order-aware Interactive Segmentation
- SegGPT: Segmenting Everything In Context
- Track Anything: Segment Anything Meets Videos
- Depth Anything V2
- SAMPart3D: Segment Any Part in 3D Objects
- OmniPart: Part-Aware 3D Generation with Semantic Decoupling and Structural Cohesion
- Concerto: Joint 2D-3D Self-Supervised Learning Emerges Spatial Representations
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models