Adding a learnable pixel-level perturbation to the in-context pair improves MAE-VQGAN visual in-context learning on segmentation and object detection benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
E-InMeMo: Enhanced Prompting for Visual In-Context Learning
Adding a learnable pixel-level perturbation to the in-context pair improves MAE-VQGAN visual in-context learning on segmentation and object detection benchmarks.