5 papers · 1 filter
Enabling Fully Integer-Only Inference for Lightweight Detection Transformers
Thanh Cong Le, Michal Szczepanski, Martyna Poreba
Vision Transformer detectors now approach the accuracy of CNNs but remain difficult to deploy on NPUs and microcontrollers because key components, including deformable attention, f…
Beyond Attention Scores: SVD-Based Vision Token Pruning for Efficient Vision-Language Models
Yvon Apedo, Martyna Poreba, Michal Szczepanski +1
Vision-Language Models (VLMs) have revolutionized multi-modal learning by jointly processing visual and textual information. Yet, they face significant challenges due to the high c…
I-Segmenter: Integer-Only Vision Transformer for Efficient Semantic Segmentation
Jordan Sassoon, Michal Szczepanski, Martyna Poreba
Vision Transformers (ViTs) have recently achieved strong results in semantic segmentation, yet their deployment on resource-constrained devices remains limited due to their high me…
Token-Space Mask Prediction for Efficient Vision Transformer Segmentation
Calvin Galagain, Martyna Poreba, François Goulette
Query-based Vision Transformer segmentation models typically reconstruct dense spatial feature maps to predict masks, inheriting design patterns from convolutional architectures. W…
Where Do Tokens Go? Understanding Pruning Behaviors in STEP at High Resolutions
Michal Szczepanski, Martyna Poreba, Karim Haroun
Vision Transformers (ViTs) achieve state-of-the-art performance in semantic segmentation but are hindered by high computational and memory costs. To address this, we propose STEP (…