7 citations · 11 across the 5 of their papers we have counts for
3 papers · 1 filter
MAEDAY: MAE for few and zero shot AnomalY-Detection
Eli Schwartz, Assaf Arbelle, Leonid Karlinsky +4
We propose using Masked Auto-Encoder (MAE), a transformer model self-supervisedly trained on image inpainting, for anomaly detection (AD). Assuming anomalous regions are harder to…
PIP: Positional-encoding Image Prior
Nimrod Shabtay, Eli Schwartz, Raja Giryes
In Deep Image Prior (DIP), a Convolutional Neural Network (CNN) is fitted to map a latent space to a degraded (e.g. noisy) image but in the process learns to reconstruct the clean…
Teaching Structured Vision&Language Concepts to Vision&Language Models
Sivan Doveh, Assaf Arbelle, Sivan Harary +8
Vision and Language (VL) models have demonstrated remarkable zero-shot performance in a variety of tasks. However, some aspects of complex language understanding still remain a cha…