BTHA: A Backbone-Transferable Adapter for Text-Guided Medical Segmentation
Researchers propose BTHA, a hierarchical adapter framework that decouples language guidance from vision and text backbones in text-guided medical image segmentation. BTHA uses a stable feature-level interface to enable reuse of language modules across heterogeneous encoder pairs without network redesign. This addresses a key limitation of existing tightly coupled architectures.
Entities
Related
TCLA: Training-Free Few-Shot Adaptation for Medical VLMs
Researchers propose TCLA, a training-free method to adapt medical Vision-Language Models (VLMs) to out-of-distribution data using only a few examples. It corrects inference logits without additional trainable components, improving stability in low-data regimes like 1-shot. The method is model-agnostic and fast, addressing domain shifts and class bias in medical imaging.
Paper challenges text-only pretraining, proposes visual pretraining for language models
A new arXiv paper argues that current language model pretraining discards rich visual information from documents and web pages. The authors propose scalable visual pretraining to incorporate figures, equations, and layouts, aiming to improve language intelligence beyond text-only approaches.
ALICE: Multi-stage distillation unifies eight pathology foundation models into one backbone
Researchers introduce ALICE, a unified pathology foundation model trained via multi-stage agglomerative distillation from eight teacher models spanning vision-only, vision-language, and slide-level expertise. Pretrained on 24,985,184 tile-level and 155,604 high-resolution images, ALICE consolidates fragmented capabilities into a single backbone and is evaluated across 21 task scenarios.
New paper proposes LLM-GCN hybrid for semi-supervised image classification
A new arXiv paper introduces a method that integrates Large Language Models with Graph Convolutional Networks to improve semi-supervised image classification. The approach addresses the challenge of graph construction for visual data by leveraging LLMs to generate better graph representations, potentially reducing the need for labeled datasets.
CLAP paper proposes direct VLM-to-VLA adaptation to isolate semantic transfer
A new arXiv paper introduces CLAP, a method that converts pretrained vision-language models (VLMs) into vision-language-action models (VLAs) with minimal architectural changes, avoiding the distribution mismatch caused by predicting actions as bare numeric tokens. The approach aims to clarify how VLM capabilities transfer across model scales for robot control.