4 papers
Coverage Matters: MarginMerge for Compressing Multi-Vector Visual Document Retrievers
Ailar Mahdizadeh, Aria Salari, Sohail Rajabi +3
Multi-vector visual document retrievers such as ColPali and ColQwen achieve strong retrieval by storing fine-grained patch embeddings, but this produces large indexes and costly la…
When Can Test-Time Adaptation Help Zero-Shot CT Vision-Language Models?
Ailar Mahdizadeh, Puria Azadi Moghadam, Xiangteng He +1
3D CT vision-language models (VLMs) classify abnormalities from text prompts in a zero-shot manner, enabling cross-institution deployment where labels are scarce and clinical tasks…
CoRDS: Coreset-based Representative and Diverse Selection for Streaming Video Understanding
Ailar Mahdizadeh, Puria Azadi, Muchen Li +2
Streaming video understanding with large vision-language models (VLMs) requires a compact memory that can support future reasoning over an ever-growing visual history. A common sol…
SCALE-VLP: Soft-Weighted Contrastive Volumetric Vision-Language Pre-training with Spatial-Knowledge Semantics
Ailar Mahdizadeh, Puria Azadi Moghadam, Xiangteng He +3
Vision-language models (VLMs) have demonstrated strong cross-modal capabilities, yet most work remains limited to 2D data and assumes binary supervision (i.e., positive vs. negativ…