Skip to main content

Table 2 Instance-level results

From: Pedestrian segmentation based on a spatio-temporally consistent graph-cut with optimal transport

 TUD-Stadtmitte
 M. IoU [%]W. IoU [%]M. IoU B [%]Time [min]
Ochs’s13.1925.963.391255
Milan’s50.9744.8020.70209
He’s70.3679.2120.230.5
Proposed57.4881.1220.4417.7
 TUD-Campus
 M. IoU [%]W. IoU [%]M. IoU B[%]Time [min]
Ochs’s15.0834.872.85500
Milan’s51.3849.2914.2783
He’s63.9271.5413.040.2
Proposed48.0877.8916.859.6
 TUD-Crossing
 M. IoU [%]W. IoU [%]M. IoU B [%]Time [min]
Ochs’s6.526.651.501512
Milan’s14.3022.875.89240
He’s38.0056.1710.080.65
Proposed30.8364.1813.5420.5
 PETS2009
 M. IoU [%]W. IoU [%]M. IoU B [%]Time [min]
Ochs’s14.4129.545.404355
Milan’s33.1734.392.24870
He’s79.2585.6142.112.1
Proposed68.2080.6137.75118
  1. The best performance data were italicized