Dynamic Multiscale Feature Fusion Method for Underwater Target Recognition

Table 4

Underwater distorted image recognition accuracy and recognition time.

Method	Torpedo	Torpedowake	Submarine	Frogman	Bubble	AUV	mAP	Time
FISHnet	0.5241	0.5872	0.7472	0.7228	0.7317	0.8380	0.6918	0.399
SiamFPN	0.5016	0.4545	0.8983	0.7150	0.6540	0.9421	0.6943	0.229
SA-FPN	0.5353	0.5450	0.7509	0.7149	0.6962	0.9180	0.6934	0.244
Literature [33]	0.5124	0.4754	0.7696	0.7583	0.7245	0.8962	0.6894	0.101
Ours	0.5554	0.4378	0.8251	0.7312	0.7854	0.9134	0.7081	0.222

Note: the bold value in the table indicates excellent indicators of each algorithm.