TDFN classifies images by processing a low-resolution global view plus a few task-selected high-resolution crops, reaching about 97.8% on MNIST.
Ehinger, Barbara Hidalgo-Sotelo , Antonio Torralba, and Aude Oliva
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Task-Driven Fixation Network: An Efficient Architecture with Fixation Selection
TDFN classifies images by processing a low-resolution global view plus a few task-selected high-resolution crops, reaching about 97.8% on MNIST.