EGRNet reports 65.28% Cityscapes mIoU with 0.46M parameters using depthwise separable convolutions, dilated residual blocks, SE attention, and an edge-gated refinement module, plus an L2-norm activation detector for adversarial inputs.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
EGRNet: A Lightweight Semantic Segmentation Network with Edge-Gated Refinement and Adversarial Sensing
EGRNet reports 65.28% Cityscapes mIoU with 0.46M parameters using depthwise separable convolutions, dilated residual blocks, SE attention, and an edge-gated refinement module, plus an L2-norm activation detector for adversarial inputs.