Table 2.

True positive rate of classification for all IPB configurations, where Multi-scale is denoted as MS, and Multi-label is denoted as ML.

Config./Class (std.)	Debris	Dense	Diff	Spread	Average
VGG19	0.8877 (0.0174)	0.7772 (0.0308)	0.8434 (0.0396)	0.9299 (0.0134)	0.8595
HRNet	0.8670 (0.0087)	0.7752 (0.0291)	0.8162 (0.0569)	0.9239 (0.0076)	0.8455
MS HRNet	0.7921 (0.0491)	0.8410 (0.0116)	0.9013 (0.0595)	0.8813 (0.0111)	0.8539
IPB VGG	0.8389 (0.0615)	0.7695 (0.0590)	0.7971 (0.0621)	0.9281 (0.0277)	0.8334
MS IPB VGG	0.8843 (0.0576)	0.8489 (0.0344)	0.8369 (0.0913)	0.8949 (0.0340)	0.8662
ML IPB VGG	0.8890 (0.0356)	0.8061 (0.0467)	0.8516 (0.0605)	0.9144 (0.0190)	0.8647
ML+MS IPB VGG	0.8905 (0.0419)	0.8529 (0.0312)	0.8259 (0.0741)	0.9006 (0.0148)	0.8674
MS IPB HRNet	0.8934 (0.0512)	0.8241 (0.0295)	0.8846 (0.0826)	0.9112 (0.0174)	0.8783
ML IPB HRNet	0.9139 (0.0364)	0.7920 (0.0311)	0.8737 (0.0785)	0.9139 (0.0158)	0.8733
ML+MS IPB HRNet	0.8843 (0.0159)	0.8575 (0.0218)	0.9085 (0.0189)	0.8985 (0.0074)	0.8872

Bold values represent best result for each column.