An Encoder-Decoder Network Based FCN Architecture for Semantic Segmentation

Table 3

Our highest scoring entry in each column is shown in italic. Results in a performance of 85.6% on PASCAL VOC 2012 test set.

Category	FCN-8s	ResSegNet	RefineNet	PSPNet	DeepLabv3+	Ours
Bicycle	34.2	65.2	73.2	72.7	77.1	78.2
Chair	21.4	37.4	43.7	43.1	56.9	57.1
Sheep	72.4	85.9	92.9	94.4	92.9	94.4
Mean	62.2	80.4	84.2	85.4	87.8	85.6