[Retracted] Rethinking Separable Convolutional Encoders for End-to-End Semantic Image Segmentation

Table 2

Performance comparison before and after improvement.

Data set	Method	Mean IoU	Mean ACC	Overall ACC
CamVid	SegProNet	0.4926	0.6129	0.8107
	Improving SegProNet	0.5013	0.6194	0.8241
VOC2012	SegProNet	0.5114	0.6237	0.8134
	Improving SegProNet	0.5207	0.6321	0.8327
CityScapes	SegProNet	0.5126	0.5904	0.8406
	Improving SegProNet	0.5225	0.6079	0.8671