Open Access

Table 2

IoU indicators for various targets under different models (unit:%)

Target category FCN-8s SegNet algorithm Algorithm in this paper(without CRF) Algorithm in this paper
road 93.2 95.1 95.0 95.2
sidewalk 80.3 80.2 80.2 80.4
building 86.3 87.9 87.5 88.1
wall 56.8 56.8 56.9 56.8
fence 40.1 43.6 43.0 44.3
pole 27.1 28.2 27.9 28.5
traffic light 37.3 37.4 37.7 37.7
traffic sign 51.1 52.6 53.0 53.1
vegetation 90.5 92.5 92.6 92.9
terrain 71.0 76.9 77.0 77.6
sky 95.2 96.3 96.7 97.0
person 52.1 50.7 50.9 51.3
rider 55.4 56.5 56.8 57.1
car 87.9 92.0 92.1 92.3
truck 45.0 47.3 47.0 48.1
bus 50.5 59.2 59.5 60.5
train 43.9 42.0 42.4 42.9
motorcycle 41.8 40.2 40.6 41.9
bicycle 54.6 58.3 58.7 59.1

Note:   The bold number indicates the most effective indicator among all comparative experiments

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.