Gastric polyp detection module based on improved attentional feature fusion

Table 2 Comparison with other attention modules

Method	MAP@0.5 (%)	F1	P (%)	R (%)
Yolov4+our feature fusion model	91.52	0.87	94.94	80.07
Yolov4+our feature fusion model with SE [17]	92.17	0.89	93.65	83.99
Yolov4+our feature fusion model with CBAM [19]	91.05	0.87	93.44	81.14
Yolov4+our feature fusion model with SRM [20]	90.98	0.88	94.67	82.17
Yolov4+our feature fusion model with SK [21]	91.70	0.89	96.25	82.21
Yolov4+our feature fusion model with ECA [22]	92.20	0.88	95.06	82.17
Yolov4+our feature fusion model with SA [23]	91.23	0.88	93.60	83.27
Yolov4+our feature fusion model with CA [18]	91.69	0.88	93.20	82.92
Yolov4+our feature fusion model with TA [24]	91.72	0.88	94.67	82.21
Yolov4+our feature fusion model with SGE [25]	92.21	0.88	94.65	81.85
Yolov4+our feature fusion model with GAM [26]	91.71	0.88	94.63	81.49
Yolov4+our attentional feature fusion model	92.33	0.89	95.92	83.63

ISSN: 1475-925X