Lightweight Real-Time Image Semantic Segmentation Network Based on Multi-Resolution Hybrid Attention Mechanism

Table 2

Setting up the dilated convolution at different stages of the backbone network, we validate on the Cityscapes set to obtain mIoU and FPS results.

Stage4	Stage3	Stage2	mIoU
—	—	—	71.1645
✓	—	—	71.0996
—	✓	—	68.9519
—	—	✓	69.0471
✓	✓	—	70.4554
✓	✓	✓	69.4500
—	✓	✓	69.4516
✓	✓	✓	71.8686