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

Table 4

Comparative experimental results (CamVid test set).

Method	Mean IoU	Mean ACC	Overall ACC
FCN-8S	0.4137	0.6217	0.6532
FCN-32S	0.4011	0.6349	0.6417
SegNet	0.4231	0.6401	0.6628
SegNet + confrontational learning	0.4327	0.6513	0.6803
SegProNet	0.4421	0.6670	0.6947
SegProNet + confrontational learning	0.4532	0.6721	0.7081
Improvement of SegProNet	0.4670	0.6739	0.7269
Improvement of SegProNet + confrontational learning	0.5730	0.6845	0.7531