A probabilistic, component-wise spatial transformer using a shared frozen tokenizer and augmentation-guided alignment loss improves geometric robustness in fine-grained moth classification.
Auto- Encoding Variational Bayes
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
-
Geometrically Constrained and Token-Based Probabilistic Spatial Transformers
A probabilistic, component-wise spatial transformer using a shared frozen tokenizer and augmentation-guided alignment loss improves geometric robustness in fine-grained moth classification.