Readout Blindness: VLM Scores Miss the Spatial Direction Their Frozen Encoders Retain
cs.CV, cs.LG
Submitted: 2026-10-05
Updated: 2026-10-05
Code: https://github.com/GarryLarry010131/ADR
Sources
- The Spatial Blindspot of Vision-Language Models
- Qwen3-VL Technical Report
- CLoVe: Encoding Compositional Language in Contrastive Vision-Language Models
- What's left can't be right -- The remaining positional incompetence of contrastive vision-language models
- SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features
- Bridging Vision and Language Concepts through Optimal Transport Semantic Flow
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