Semantic Understandings for Aerial Images via Multigrained Feature Grouping

Table 2

Performance comparison with state-of-the-art methods on the UCM multilabel dataset (%).

Model	EP	ER	EF1	EF2	LP	LR	LF1	LF2
ReNet-50 [40]	80.86	81.95	81.4	81.73	88.78	78.98	83.59	80.76
ResNet-RBFNN [41]	79.92	84.59	82.19	83.61	86.21	83.72	84.95	84.21
CA-ResNet-LSTM [15]	79.9	86.14	82.90	84.82	86.99	82.24	84.55	83.15
CA-ResNet-BiLSTM [15]	77.94	89.02	83.11	86.56	86.12	84.26	85.18	84.63
Image_GCN [42]	75.00	69.00	71.86	70.12	76.00	69.00	72.33	70.29
ML_GCN [34]	79.86	82.10	80.96	81.64	86.42	80.83	83.53	81.89
MSGM	83.86	85.48	84.54	85.10	89.98	85.07	87.46	86.01