DSSINet combines CRF-based multi-scale feature refinement and a dilated MS-SSIM loss to improve crowd counting accuracy on four benchmarks.
Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolu- tion, and fully connected crfs
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
-
Crowd Counting with Deep Structured Scale Integration Network
DSSINet combines CRF-based multi-scale feature refinement and a dilated MS-SSIM loss to improve crowd counting accuracy on four benchmarks.