Standard convolutional networks can classify tiny objects in large images only above a certain object-to-image ratio, and the training data needed to reach that point rises rapidly as the object gets smaller.
Multiple object recognition with visual attention
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Needles in Haystacks: On Classifying Tiny Objects in Large Images
Standard convolutional networks can classify tiny objects in large images only above a certain object-to-image ratio, and the training data needed to reach that point rises rapidly as the object gets smaller.