Skip to main content

Table 2 Median of differences between different models

From: Semantic segmentation and photogrammetry of crowdsourced images to monitor historic facades

  IoU (%) F1 (%) MPA (%) Precision (%) Recall (%)
Overall    
 M1–M2 14.6 (5.3, 25.6) 15.3 (5.0, 25.4) 1.1 (− 3.1, 7.0) 14.4 (4.6, 27.5) 13.9 (1.0, 17.6)
 M1–M3 − 2.3 (− 5.1, 0.3) − 2.3 (− 3.5, 0.0) − 1.1 (− 3.9, 0.6) − 2.7 (− 3.6, 0.3) − 1.4 (− 4.2, 0.6)
 M2–M3 − 21.3 (− 26.1, − 9.7) − 20.7 (− 26.6, − 7.7) − 2.0 (− 6.3, 1.3) − 21.2 (− 29.1, − 9.0) − 12.9 (− 19.2, − 2.5)
Plant (1s)  
 M1–M2 − 0.9 (− 11.8, 8.5) − 0.8 (− 10.7, 8.3) 0.7 (− 2.4, 2.9) − 6.5 (− 13.6, 10.0) 2.3 (− 4.6, 6.0)
 M1–M3 − 8.8 (− 14.6, − 1.0) − 9.0 (− 16.0, − 0.8) − 1.7 (− 3.8, − 0.2) − 7.8 (− 14.8, 0.4) − 3.3 (− 8.7, − 0.1)
 M2–M3 − 6.2 (− 14.7, 3.0) − 4.9 (− 16.2, 2.4) − 0.7 (− 5.5, 0.2) 2.7 (− 12.5, 5.6) − 2.4 (− 12.5, 0.1)
Plant (2s)  
 M1–M2 10.9 (6.5, 25.2) 9.6 (6.7, 27.2) 0.1 (− 0.3, 1.1) 13.4 (7.4, 22.1) 0.2 (− 0.4, 1.9)
 M1–M3 2.9 (− 3.7, 5.7) 2.2 (− 3.5, 4.7) − 0.1 (− 1.4, 0.5) 1.5 (− 3.8, 5.5) − 0.2 (− 3.1, 1.1)
 M2–M3 − 10.5 (− 26.1, − 2.6) − 10.2 (− 29.7, − 2.6) − 0.3 (− 1.0, 0.1) − 10.4 (− 23.1, − 2.9) − 0.4 (− 1.9, − 0.0)
  1. The numbers in brackets are the 25th and 75th percentiles