A feature refinement module that aggregates multi-stage features with non-local attention improves light-weight semantic segmentation, reaching 80.4% mIoU on Cityscapes at 214.82 GFLOPs.
Specifi- cally, FRM firstly pools the multi-stage features with differ- ent scales to the same size H 32 × W 32 and concatenates them together
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A feature refinement module for light-weight semantic segmentation network
A feature refinement module that aggregates multi-stage features with non-local attention improves light-weight semantic segmentation, reaching 80.4% mIoU on Cityscapes at 214.82 GFLOPs.