Fast and Accurate Scene Parsing via Bi-direction Alignment Networks

Open paper

Benchmark	Model	Rank	Results
semantic-segmentation-on-bdd100k-val	Bi-Align	#18	mIoU: 53.4(42.1fps)