A text-conditioned YOLOv3 variant with spatial features directly regresses a grounding box and outperforms propose-and-rank two-stage methods on Flickr30K Entities and ReferItGame while being about 10x faster.
The pascal visual object classes (voc) challenge
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
-
A Fast and Accurate One-Stage Approach to Visual Grounding
A text-conditioned YOLOv3 variant with spatial features directly regresses a grounding box and outperforms propose-and-rank two-stage methods on Flickr30K Entities and ReferItGame while being about 10x faster.