Vision-Language Models (VLMs) and Vision-Language-Action (VLA) models process hundreds of visual tokens, making visual representations a major computational bottleneck. We argue that visual token pruning should be formulated not as identifying individually important tokens, but as preserving the dominant low-rank spect...
Yvon Apedo, Martyna Poreba, M. Szczepanski et al.· 0 citations
Vision Transformers (ViTs) have achieved state-of-the-art performance across a range of computer vision tasks, mainly thanks to the self-attention mechanism. However, its complexity, increasing quadratically with the number of tokens, remains the major obstacle to ViT efficiency and deployment at scale. Token merging r...
Victor Bercy, Martyna Poreba, M. Szczepanski et al.· arXiv.org· 0 citations
Results show that segmentation-region dynamicity can be learned from propagated queries, enabling its online prediction without a dedicated motion-processing pipeline at inference, and shows that segmentation-region dynamicity can be learned from propagate queries.
Calvin Galagain, Martyna Poreba, François Goulette· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.